1 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
3 * config.gcc (extra_headers): Add loongson-mmiintrin.h.
4 * config/mips/loongson.md: Move to ...
5 * config/mips/loongson-mmi.md: here; Adjustment.
6 * config/mips/loongson.h: Move to ...
7 State as deprecated. Include loongson-mmiintrin.h for back
8 compatibility and warning.
9 * config/mips/loongson-mmiintrin.h: ... here.
10 * config/mips/mips.c (mips_hard_regno_mode_ok_uncached,
11 mips_vector_mode_supported_p, AVAIL_NON_MIPS16): Use
12 TARGET_LOONGSON_MMI instead of TARGET_LOONGSON_VECTORS.
13 (mips_option_override): Make sure MMI use hard float;
14 (mips_shift_truncation_mask, mips_expand_vpc_loongson_even_odd,
15 mips_expand_vpc_loongson_pshufh, mips_expand_vpc_loongson_bcast,
16 mips_expand_vector_init): Use TARGET_LOONGSON_MMI instead of
17 TARGET_LOONGSON_VECTORS.
18 * gcc/config/mips/mips.h (TARGET_LOONGSON_VECTORS): Delete.
19 (TARGET_CPU_CPP_BUILTINS): Add __mips_loongson_mmi.
20 (MIPS_ASE_DSP_SPEC, MIPS_ASE_LOONGSON_MMI_SPEC): New.
21 (BASE_DRIVER_SELF_SPECS): march=loongson2e/2f/3a implies
23 (SHIFT_COUNT_TRUNCATED): Use TARGET_LOONGSON_MMI instead of
24 TARGET_LOONGSON_VECTORS.
25 * gcc/config/mips/mips.md (MOVE64, MOVE128): Use
26 TARGET_LOONGSON_MMI instead of TARGET_LOONGSON_VECTORS.
27 (Loongson MMI patterns): Include loongson-mmi.md instead of
29 * gcc/config/mips/mips.opt (-mloongson-mmi): New option.
30 * gcc/doc/invoke.texi (-mloongson-mmi): Document.
32 2018-11-07 Richard Biener <rguenther@suse.de>
35 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Fixup
36 BLOCK_ABSTRACT_ORIGIN to be the ultimate origin.
38 2018-11-07 Alexandre Oliva <aoliva@redhat.com>
40 PR rtl-optimization/87874
41 * lra.c (lra_substitute_pseudo): Do not create a subreg for
44 2018-11-06 Aaron Sawdey <acsawdey@linux.ibm.com>
46 * config/rs6000/rs6000.md (bswap<mode>2): Force address into register
47 if not in indexed or indirect form.
48 (bswap<mode>2_load): Change predicate to indexed_or_indirect_operand.
49 (bswap<mode>2_store): Ditto.
51 2018-11-06 Richard Earnshaw <rearnsha@arm.com>
53 * config/aarch64/aarch64.md (speculation_tracker): Set the mode for
56 2018-11-06 Richard Biener <rguenther@suse.de>
58 PR tree-optimization/86850
59 * vec.h (vec<T, va_heap, vl_ptr>::splice): Check src.length ()
62 2018-11-06 Jan Hubicka <jh@suse.cz>
64 * tree.c (fld_simplified_type_name): Break out form ...
65 (free_lang_data_in_type): ... here.
66 (fld_type_variant_equal_p): Use it.
68 2018-11-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
70 * config/default-d.c: Include memmodel.h.
72 * config/sol2-d.c: New file.
73 * config/t-sol2 (sol2-d.o): New rule.
74 * config.gcc <*-*-solaris2*>: Set d_target_objs,
77 2018-11-06 Jan Hubicka <jh@suse.cz>
79 * tree.c (fld_type_variant): Also copy alignment; be sure that
82 2018-11-06 Ilya Leoshkevich <iii@linux.ibm.com>
85 * config/s390/s390.md: Add relative_long attribute.
87 2018-11-06 Jan Hubicka <jh@suse.cz>
89 * ipa-pure-const.c (check_decl): Do not test TYPE_NEEDS_CONSTRUCTING.
90 * lto-streamer-out.c (hash_tree): Do not hash TYPE_NEEDS_CONSTRUCTING.
91 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Do not
92 stream TYPE_NEEDS_CONSTRUCTING.
93 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
94 * tree.c (free_lang_data_in_type): Clear TYPE_NEEDS_CONSTRUCTING.
96 2018-11-06 Richard Biener <rguenther@suse.de>
98 * tree-vect-slp.c (vect_slp_bb): Move opening of vect_slp_analyze_bb
100 (vect_slp_analyze_bb_1): ... here to avoid hiding optimized locations.
102 2018-11-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
104 * gcc/config/msp430/msp430.h (REG_CLASS_CONTENTS): Add R0 to
105 REG_CLASS_CONTENTS[GEN_REGS].
106 (REGNO_REG_CLASS): Return NO_REGS for R2 and R3.
108 2018-11-06 Jan Hubicka <jh@suse.cz>
110 * tree.c (fld_simplified_type_of): Clear TYPELESS_STORAGE flag.
112 2018-11-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
115 * config/sol2.h (ASAN_CC1_SPEC): Define.
116 (LD_WHOLE_ARCHIVE_OPTION): Define.
117 (LD_NO_WHOLE_ARCHIVE_OPTION): Define.
118 (ASAN_REJECT_SPEC): Provide default.
119 (LIBASAN_EARLY_SPEC): Define.
120 (LIBTSAN_EARLY_SPEC): Define.
121 (LIBLSAN_EARLY_SPEC): Define.
122 * config/i386/sol2.h (CC1_SPEC): Redefine.
123 (ASAN_REJECT_SPEC): Define.
125 * config/sparc/sparc.c (sparc_asan_shadow_offset): Declare.
126 (TARGET_ASAN_SHADOW_OFFSET): Define.
127 (sparc_asan_shadow_offset): New function.
128 * config/sparc/sol2.h (CC1_SPEC): Append ASAN_CC1_SPEC.
129 (ASAN_REJECT_SPEC): Define.
131 2018-11-06 Jan Hubicka <jh@suse.cz>
133 * tree.c (fld_type_variant): Copy canonical type.
134 (fld_incomplete_type_of): Check that canonical types looks sane;
136 (verify_type): Accept when incomplete type has complete canonical type.
138 2018-11-06 Jan Hubicka <jh@suse.cz>
140 * tree.c (free_lang_data): Reset overwite_assembler_name,
141 print_xnode, print_decl, print_type and print_identifier of
144 2018-11-06 Richard Biener <rguenther@suse.de>
146 PR tree-optimization/87889
147 * tree-vect-loop-manip.c (slpeel_duplicate_current_defs_from_edges):
148 Do nothing if old and new arg are the same
150 2018-11-06 Andreas Krebbel <krebbel@linux.ibm.com>
153 * config/s390/s390.md ("*r<noxa>sbg_di_rotl"): Remove mode
154 attributes for operands 3 and 4.
156 2018-11-06 Richard Biener <rguenther@suse.de>
159 * simplify-rtx.c (simplify_binary_operation_1): Add pattern
160 matching bitfield insertion.
162 2018-11-06 Alexandre Oliva <aoliva@redhat.com>
164 * auto-inc-dec.c: Include valtrack.h. Improve comments.
165 (reg_next_debug_use): New.
166 (attempt_change): Propagate adjusted expression into affected
168 (merge_in_block): Track uses in debug insns.
169 (pass_inc_dec::execute): Allocate and release
172 2018-11-06 Wei Xiao <wei3.xiao@intel.com>
174 * config/i386/avx512fintrin.h: Update VFIXUPIMM* intrinsics.
175 (_mm512_fixupimm_round_pd): Update parameters and builtin.
176 (_mm512_maskz_fixupimm_round_pd): Ditto.
177 (_mm512_fixupimm_round_ps): Ditto.
178 (_mm512_maskz_fixupimm_round_ps): Ditto.
179 (_mm_fixupimm_round_sd): Ditto.
180 (_mm_maskz_fixupimm_round_sd): Ditto.
181 (_mm_fixupimm_round_ss): Ditto.
182 (_mm_maskz_fixupimm_round_ss): Ditto.
183 (_mm512_fixupimm_pd): Ditto.
184 (_mm512_maskz_fixupimm_pd): Ditto.
185 (_mm512_fixupimm_ps): Ditto.
186 (_mm512_maskz_fixupimm_ps): Ditto.
187 (_mm_fixupimm_sd): Ditto.
188 (_mm_maskz_fixupimm_sd): Ditto.
189 (_mm_fixupimm_ss): Ditto.
190 (_mm_maskz_fixupimm_ss): Ditto.
191 (_mm512_mask_fixupimm_round_pd): Update builtin.
192 (_mm512_mask_fixupimm_round_ps): Ditto.
193 (_mm_mask_fixupimm_round_sd): Ditto.
194 (_mm_mask_fixupimm_round_ss): Ditto.
195 (_mm512_mask_fixupimm_pd): Ditto.
196 (_mm512_mask_fixupimm_ps): Ditto.
197 (_mm_mask_fixupimm_sd): Ditto.
198 (_mm_mask_fixupimm_ss): Ditto.
199 * config/i386/avx512vlintrin.h:
200 (_mm256_fixupimm_pd): Update parameters and builtin.
201 (_mm256_maskz_fixupimm_pd): Ditto.
202 (_mm256_fixupimm_ps): Ditto.
203 (_mm256_maskz_fixupimm_ps): Ditto.
204 (_mm_fixupimm_pd): Ditto.
205 (_mm_maskz_fixupimm_pd): Ditto.
206 (_mm_fixupimm_ps): Ditto.
207 (_mm_maskz_fixupimm_ps): Ditto.
208 (_mm256_mask_fixupimm_pd): Update builtin.
209 (_mm256_mask_fixupimm_ps): Ditto.
210 (_mm_mask_fixupimm_pd): Ditto.
211 (_mm_mask_fixupimm_ps): Ditto.
212 * config/i386/i386-builtin-types.def: Add new types and remove useless ones.
213 * config/i386/i386-builtin.def: Update builtin definitions.
214 * config/i386/i386.c: Handle new builtin types and remove useless ones.
215 * config/i386/sse.md: Update VFIXUPIMM* patterns.
216 (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
217 (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
218 (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Update.
219 (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
220 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
221 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Update.
222 * config/i386/subst.md:
223 (round_saeonly_sd_mask_operand4): Add new subst_attr.
224 (round_saeonly_sd_mask_op4): Ditto.
225 (round_saeonly_expand_operand5): Ditto.
226 (round_saeonly_expand): Update.
228 2018-11-05 Max Filippov <jcmvbkbc@gmail.com>
230 * config/xtensa/uclinux.h (XTENSA_ALWAYS_PIC): Change to 0.
232 2018-11-05 Segher Boessenkool <segher@kernel.crashing.org>
234 PR rtl-optimization/87871
235 * combine.c (make_more_copies): Skip if dest is frame_pointer_rtx.
237 2018-11-05 Paul Koning <ni1d@arrl.net>
239 * doc/sourcebuild.texi (target attributes): Document new "inf"
240 effective target keyword.
242 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
244 * config/s390/s390.c (s390_register_move_cost): Increase costs for
245 moves involving the CC reg.
247 2018-11-05 Richard Biener <rguenther@suse.de>
249 PR tree-optimization/87873
250 * tree-ssa-loop-manip.h (split_loop_exit_edge): Add copy_constants_p
252 * tree-ssa-loop-manip.c (split_loop_exit_edge): Likewise.
253 * tree-vect-loop.c (vect_transform_loop): When splitting the
254 loop exit also create forwarder PHIs for constants.
255 * tree-vect-loop-manip.c (slpeel_duplicate_current_defs_from_edges):
256 Handle constant to_arg, add extra checking we match up the correct
259 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
261 * config/s390/s390.md: QImode and HImode for load on condition.
263 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
265 * config/s390/predicates.md: Fix typo.
266 * config/s390/s390.md: Allow immediates for load on condition.
268 2018-11-05 Martin Liska <mliska@suse.cz>
270 * alloc-pool.h (struct pool_usage): Use SIZE_AMOUNT.
271 * bitmap.h (struct bitmap_usage): Likewise.
272 * ggc-common.c (SCALE): Remove.
274 (struct ggc_usage): Use SIZE_AMOUNT. And update
276 * ggc-page.c (SCALE): Remove.
277 (STAT_LABEL): Remove.
278 (ggc_print_statistics): Use SIZE_AMOUNT.
279 * gimple.h (SCALE): Remove.
281 * input.c (ONE_K): Remove.
284 (STAT_LABEL): Likewise.
285 (FORMAT_AMOUNT): Likewise.
286 (dump_line_table_statistics): Use SIZE_AMOUNT.
287 * mem-stats.h (struct mem_usage): Likewise.
288 * rtl.c (dump_rtx_statistics): Likewise.
289 (rtx_alloc_counts): Change type to size_t.
290 (rtx_alloc_sizes): Likewise.
291 (rtx_count_cmp): New.
292 (dump_rtx_statistics): Sort first based on counts.
293 * tree.c (tree_nodes_cmp): New.
294 (tree_codes_cmp): New.
295 (dump_tree_statistics): Sort first based on counts.
296 * system.h (ONE_K): New.
298 (SIZE_SCALE): Likewise.
299 (SIZE_LABEL): Likewise.
300 (SIZE_AMOUNT): Likewise.
301 * tree-cfg.c (dump_cfg_stats): Use SIZE_AMOUNT.
302 * tree-dfa.c (dump_dfa_stats): Likewise.
303 * tree-phinodes.c (phinodes_print_statistics): Likewise.
304 * tree-ssanames.c (ssanames_print_statistics): Likewise.
305 * tree.c (dump_tree_statistics): Likewise.
306 * vec.c (struct vec_usage): Likewise.
307 * trans-mem.c (tm_mangle): Enlarge buffer in order to not
308 trigger a -Werror=format-overflow with
309 --enable-gather-detailed-stats.
311 2018-11-05 Martin Liska <mliska@suse.cz>
313 * mem-stats.h (mem_alloc_description::release_instance_overhead):
315 * vec.c (struct vec_usage): Register m_element_size.
316 (vec_prefix::register_overhead): New arguments: elements and
318 (vec_prefix::release_overhead): Subtract elements.
319 * vec.h (struct vec_prefix): Change signature.
320 (va_heap::reserve): Pass proper arguments.
321 (va_heap::release): Likewise.
323 2018-11-05 Martin Liska <mliska@suse.cz>
325 * mem-stats.h (mem_alloc_description::get_list): Fix GNU coding
329 2018-11-05 Richard Biener <rguenther@suse.de>
331 * tree-scalar-evolution.h (final_value_replacement_loop): Update
333 * tree-scalar-evolution.c (final_value_replacement_loop): Return
334 whether anything was done.
335 (scev_const_prop): Remove constant propagation part, fold
337 * tree-ssa-loop.c (pass_scev_cprop::execute): ... here.
338 (pass_data_scev_cprop): TODO_cleanup_cfg is now done
341 2018-11-05 Jakub Jelinek <jakub@redhat.com>
343 PR tree-optimization/87859
344 * gimple-ssa-store-merging.c (struct merged_store_group): Add
345 only_constants and first_nonmergeable_order members.
346 (merged_store_group::merged_store_group): Initialize them.
347 (merged_store_group::do_merge): Clear only_constants member if
348 adding something other than INTEGER_CST store.
349 (imm_store_chain_info::coalesce_immediate_stores): Don't merge
350 stores with order >= first_nonmergeable_order. Use
351 merged_store->only_constants instead of always recomputing it.
352 Set merged_store->first_nonmergeable_order if we've skipped any
353 stores. Attempt to merge overlapping INTEGER_CST stores that
354 we would otherwise skip.
357 * match.pd (X + Y < X): Don't optimize if TYPE_OVERFLOW_SANITIZED.
359 2018-11-05 Xuepeng Guo <xuepeng.guo@intel.com>
362 * config/i386/emmintrin.h (__v16qs): New to cope with option
364 (_mm_cmpeq_epi8): Replace __v16qi with __v16qs.
365 (_mm_cmplt_epi8): Likewise.
366 (_mm_cmpgt_epi8): Likewise.
368 2018-11-05 Richard Biener <rguenther@suse.de>
370 PR rtl-optimization/87852
371 * fwprop.c (use_killed_between): Only consider single-defs of the
372 use whose definition statement dominates the use.
374 2018-11-05 Martin Liska <mliska@suse.cz>
377 * doc/invoke.texi: Remove options that are
378 not disabled with -Os.
380 2018-11-05 Martin Liska <mliska@suse.cz>
383 * doc/extend.texi: Update constrain about the last argument
384 of __builtin_expect_with_probability.
386 2018-11-05 Martin Liska <mliska@suse.cz>
389 * predict.c (expr_expected_value_1): Verify
390 that last argument is a real constants and emit
393 2018-11-05 Martin Liska <mliska@suse.cz>
395 PR gcov-profile/77698
396 * ipa-profile.c (ipa_profile): Adjust hotness threshold
399 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
401 PR tree-optimization/86572
402 * builtins.c (c_strlen): Handle negative offsets in a safe way.
404 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
406 PR tree-optimization/87672
407 * gimple-fold.c (gimple_fold_builtin_stxcpy_chk): Gimplify.
408 * tree-ssa-strlen.c (handle_builtin_strcat): Adjust object size.
410 2018-11-04 Uros Bizjak <ubizjak@gmail.com>
413 * cfgexpand.c (pass_expand::execute): Move the call to
414 finish_eh_generation in front of the call to expand_stack_alignment.
416 2018-11-04 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
418 * common/config/i386/i386-common.c (processor_alias_table): Add
420 * config.gcc (i[34567]86-*-linux* | ...): Add znver2.
421 (case ${target}): Add znver2.
422 * config/i386/driver-i386.c: (host_detect_local_cpu): Let
423 -march=native recognize znver2 processors.
424 * config/i386/i386-c.c (ix86_target_macros_internal): Add znver2.
425 * config/i386/i386.c (m_znver2): New definition.
426 (m_ZNVER): New definition.
427 (m_AMD_MULTIPLE): Includes m_znver2.
428 (processor_cost_table): Add znver2 entry.
429 (processor_target_table): Add znver2 entry.
430 (get_builtin_code_for_version): Set priority for
432 (processor_model): Add M_AMDFAM17H_ZNVER2.
433 (arch_names_table): Ditto.
434 (ix86_reassociation_width): Include znver2.
435 * config/i386/i386.h (TARGET_znver2): New definition.
436 (struct ix86_size_cost): Add TARGET_ZNVER2.
437 (enum processor_type): Add PROCESSOR_ZNVER2.
438 * config/i386/i386.md (define_attr "cpu"): Add znver2.
439 * config/i386/x86-tune-costs.h: (processor_costs) Add znver2 costs.
440 * config/i386/x86-tune-sched.c: (ix86_issue_rate): Add znver2.
441 (ix86_adjust_cost): Add znver2.
442 * config/i386/x86-tune.def: Replace m_ZNVER1 by m_ZNVER.
443 * gcc/doc/extend.texi: Add details about znver2.
444 * gcc/doc/invoke.texi: Add details about znver2.
446 2018-11-03 Sandra Loosemore <sandra@codesourcery.com>
450 * config/nios2/nios2.c (nios2_rtx_costs): Recognize <mul>sidi3
453 2018-11-02 Aaron Sawdey <acsawdey@linux.ibm.com>
455 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Pay
456 attention to TARGET_AVOID_XFORM and BYTES_BIG_ENDIAN.
458 2018-11-02 Richard Earnshaw <rearnsha@arm.com>
460 * config/aarch64/aarch64.c ((aarch64_override_options): Disable
461 shrink-wrapping when -mtrack-speculation.
463 2018-11-02 Richard Biener <rguenther@suse.de>
465 * tree-ssa-coalesce.c (struct coalesce_list): Add obstack member.
466 (pop_cost_one_pair): Do not free pair.
467 (pop_best_coalesce): Likewise.
468 (create_coalesce_list): Initialize obstack.
469 (delete_coalesce_list): Free obstack.
470 (find_coalesce_pair): Obstack-allocate coalesce pairs.
471 (add_cost_one_coalesce): Likewise.
472 (struct live_track): Remove bitmap pointer indirections.
473 (new_live_track): Adjust.
474 (delete_live_track): Likewise.
475 (live_track_remove_partition): Likewise.
476 (live_track_add_partition): Likewise.
477 (live_track_live_p): Likewise.
478 (live_track_process_def): Likewise.
479 (live_track_clear_base_vars): Likewise.
481 2018-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
483 * configure.ac (gcc_cv_as_sparc_register_op): Remove.
484 * configure: Regenerate.
485 * config.in: Regenerate.
486 * config/sparc/sparc.c (sparc_output_scratch_registers): Remove
487 HAVE_AS_REGISTER_PSEUDO_OP guard.
488 * config/sparc/sparc.h (ASM_ARCH64_SPEC): Remove
489 !HAVE_AS_REGISTER_PSEUDO_OP support.
490 (ASM_DECLARE_REGISTER_GLOBAL): Remove HAVE_AS_REGISTER_PSEUDO_OP
493 2018-11-02 Richard Biener <rguenther@suse.de>
495 * dwarf2out.c (build_abbrev_table): Guard lookup_external_ref call
498 2018-11-02 Richard Biener <rguenther@suse.de>
500 PR tree-optimization/87776
501 * tree-ssa-sccvn.c (do_rpo_vn): Do not mark backedges
502 executable when iterating but running into rpo-vn-max-loop-depth
503 and not eliding the iteration.
505 2018-11-30 Jan Hubicka <jh@suse.cz>
507 * tree.c (free_lang_data_in_decl): Free TREE_TYPE and alignment of
510 2018-11-01 Aaron Sawdey <acsawdey@linux.ibm.com>
512 * config/rs6000/rs6000-protos.h (rs6000_address_for_fpconvert): Remove
514 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem):
515 Combine with rs6000_address_for_fpconvert.
516 (rs6000_address_for_fpconvert): Combine with
517 rs6000_force_indexed_or_indirect_mem.
518 (rs6000_expand_vector_init): Change function call from
519 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
520 * config/rs6000/rs6000.md (floatsi<mode>2_lfiwax): Change call from
521 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
522 (floatsi<mode>2_lfiwax_mem): Ditto.
523 (floatunssi<mode>2_lfiwzx): Ditto.
524 (floatunssi<mode>2_lfiwzx_mem): Ditto.
525 (float<QHI:mode><FP_ISA3:mode>2): Ditto.
526 (floatuns<QHI:mode><FP_ISA3:mode>2): Ditto.
527 (fix_trunc<mode>si2_stfiwx): Ditto.
528 (fixuns_trunc<mode>si2_stfiwx): Ditto.
529 (float_<mode>si2_hw): Ditto.
530 (floatuns_<mode>si2_hw): Ditto.
531 * config/rs6000/vsx.md (*vsx_extract_si): Ditto.
532 (vsx_splat_<mode>): Ditto.
534 2018-11-01 Joseph Myers <joseph@codesourcery.com>
536 * configure.ac (gcc_cv_as_ia64_ltoffx_ldxmov_relocs): Add newline
537 at end of assembler input text.
538 * configure: Regenerate.
540 2018-11-01 Jakub Jelinek <jakub@redhat.com>
542 PR tree-optimization/87826
543 * gimple-ssa-store-merging.c (do_shift_rotate): Punt if count is
544 negative or larger or equal to type's precision.
546 2018-10-31 Alexandre Oliva <aoliva@redhat.com>
548 * opts.c (default_options_table): Do not enable
549 OPT_fdelayed_branch at -Og.
550 * doc/invoke.texi (-fdelayed-branch): Document it.
552 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
554 * optabs-libfuncs.c (build_libfunc_function_visibility):
555 New, split out from...
556 (build_libfunc_function): ... here.
557 (init_one_libfunc_visibility): New, split out from ...
558 (init_one_libfunc): ... here.
560 * config/aarch64/atomics.md (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse):
561 scratch register need not be early-clobber. Document the reason
562 why we cannot use ST<OP>.
564 2018-10-31 Joseph Myers <joseph@codesourcery.com>
567 * configure.ac: Remove AC_PREREQ. Use AC_LANG_SOURCE. Use single
568 line for second argument of AC_DEFINE_UNQUOTED.
569 * doc/install.texi (Tools/packages necessary for modifying GCC):
570 Update to autoconf 2.69 and automake 1.15.1.
571 * aclocal.m4, config.in, configure: Regenerate.
573 2018-10-31 Pat Haugen <pthaugen@us.ibm.com>
575 * tree-ssa-loop-ivcanon.c (try_peel_loop): Fix typo and remove dead
578 2018-10-31 Martin Liska <mliska@suse.cz>
581 * common/config/aarch64/aarch64-common.c (aarch64_parse_extension):
582 Add new argument invalid_extension.
583 (aarch64_get_all_extension_candidates): New function.
584 (aarch64_rewrite_selected_cpu): Add NULL to function call.
585 * config/aarch64/aarch64-protos.h (aarch64_parse_extension): Add
587 (aarch64_get_all_extension_candidates): New function.
588 * config/aarch64/aarch64.c (aarch64_parse_arch): Add new
589 argument invalid_extension.
590 (aarch64_parse_cpu): Likewise.
591 (aarch64_print_hint_for_extensions): New function.
592 (aarch64_validate_mcpu): Provide hint about invalid extension.
593 (aarch64_validate_march): Likewise.
594 (aarch64_handle_attr_arch): Pass new argument.
595 (aarch64_handle_attr_cpu): Provide hint about invalid extension.
596 (aarch64_handle_attr_isa_flags): Likewise.
598 2018-10-31 Richard Biener <rguenther@suse.de>
602 * tree-outof-ssa.c (insert_backedge_copies): Restrict
603 copy generation to useful cases. Place the copy before
604 the definition of the backedge value when possible.
606 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
608 * config/arc/arc-arch.h (ARC_TUNE_ARC7XX): New tune value.
609 * config/arc/arc.c (arc_active_insn): New function.
610 (check_store_cacheline_hazard): Likewise.
611 (workaround_arc_anomaly): Use check_store_cacheline_hazard.
612 (arc_override_options): Disable delay slot scheduler for older
614 (arc_store_addr_hazard_p): New implementation, old one renamed to
616 (arc_store_addr_hazard_internal_p): Renamed.
617 (arc_reorg): Don't combine into brcc instructions which are part
618 of hardware hazard solution.
619 * config/arc/arc.md (attr tune): Consider new arc7xx tune value.
620 (tune_arc700): Likewise.
621 * config/arc/arc.opt (arc7xx): New tune value.
622 * config/arc/arc700.md: Improve A7 scheduler.
624 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
626 * config/arc/arc.c (arc_override_options): Remove
627 TARGET_COMPACT_CASESI.
628 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Update.
629 (CASE_VECTOR_MODE): Likewise.
630 (CASE_VECTOR_PC_RELATIVE): Likewise.
631 (CASE_VECTOR_SHORTEN_MODE): Likewise.
632 (CASE_VECTOR_SHORTEN_MODE1): Delete.
633 (ADDR_VEC_ALIGN): Update.
634 (ASM_OUTPUT_CASE_LABEL): Undefine.
635 (ASM_OUTPUT_BEFORE_CASE_LABEL): Undefine.
636 (TARGET_BI_BIH): Define.
637 (DEFAULT_BRANCH_INDEX): Likewise.
638 * config/arc/arc.md (casesi): Rework to accept BI/BIH
639 instructions, remove compact_casesi use case.
640 (casesi_compact_jump): Remove.
641 (casesi_dispatch): New pattern.
642 * config/arc/arc.opt: Add mbranch-index option. Deprecate
643 compact_casesi option.
644 * doc/invoke.texi: Document mbranch-index option.
646 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
648 * config/arc/arc.c (arc_get_tp): Remove function.
649 (arc_emit_call_tls_get_addr): Likewise.
650 (arc_call_tls_get_addr): New function.
651 (arc_legitimize_tls_address): Make use of arc_call_tls_get_addr.
652 * config/arc/arc.md (tls_load_tp_soft): Remove.
653 (tls_gd_get_addr): Likewise.
655 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
657 * config/arc/arc.md (mulsi3): Remove call to mulsi_600_lib.
658 (mulsi3_600_lib): Remove pattern.
659 (umulsi3_highpart_600_lib_le): Likewise.
660 (umulsi3_highpart): Remove call to umulsi3_highpart_600_lib_le.
661 (umulsidi3): Remove call to umulsidi3_600_lib.
662 (umulsidi3_600_lib): Remove pattern.
663 (peephole2): Remove peephole using the above deprecated patterns.
665 2018-10-31 Thomas Preud'homme <thomas.preudhomme@linaro.org>
668 * config/arm/arm.c (arm_option_check_internal): Disable the combined
669 use of -mslow-flash-data and -mword-relocations.
670 (arm_option_override): Enable -mword-relocations if -fpic or -fPIC.
671 * config/arm/arm.md (SYMBOL_REF MOVT splitter): Stop checking for
673 * doc/invoke.texi (-mword-relocations): Mention conflict with
675 (-mslow-flash-data): Reciprocally.
677 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
679 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Force
680 16-byte modes held in GP registers to use an even regno.
682 * config/aarch64/aarch64.c (aarch64_emit_bic): Remove.
683 (aarch64_atomic_ldop_supported_p): Remove.
684 (aarch64_gen_atomic_ldop): Remove.
685 * config/aarch64/atomic.md (atomic_<atomic_optab><ALLI>):
686 Fully expand LSE operations here.
687 (atomic_fetch_<atomic_optab><ALLI>): Likewise.
688 (atomic_<atomic_optab>_fetch<ALLI>): Likewise.
689 (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse): Drop atomic_op iterator
690 and use ATOMIC_LDOP instead; use register_operand for the input;
691 drop the split and emit insns directly.
692 (aarch64_atomic_fetch_<ATOMIC_LDOP><ALLI>_lse): Likewise.
693 (aarch64_atomic_<atomic_op>_fetch<ALLI>_lse): Remove.
694 (@aarch64_atomic_load<ATOMIC_LDOP><ALLI>): Remove.
696 * config/aarch64/aarch64.c (aarch64_emit_atomic_swap): Remove.
697 (aarch64_gen_atomic_ldop): Don't call it.
698 * config/aarch64/atomics.md (atomic_exchange<ALLI>):
699 Use aarch64_reg_or_zero.
700 (aarch64_atomic_exchange<ALLI>): Likewise.
701 (aarch64_atomic_exchange<ALLI>_lse): Remove split; remove & from
702 operand 0; use aarch64_reg_or_zero for input; merge ...
703 (@aarch64_atomic_swp<ALLI>): ... this and remove.
705 * config/aarch64/aarch64.c (aarch64_gen_compare_reg_maybe_ze): New.
706 (aarch64_split_compare_and_swap): Use it.
707 (aarch64_expand_compare_and_swap): Likewise. Remove convert_modes;
708 test oldval against the proper predicate.
709 * config/aarch64/atomics.md (@atomic_compare_and_swap<ALLI>):
710 Use nonmemory_operand for expected.
711 (cas_short_expected_pred): New.
712 (@aarch64_compare_and_swap<SHORT>): Use it; use "rn" not "rI" to match.
713 (@aarch64_compare_and_swap<GPI>): Use "rn" not "rI" for expected.
714 * config/aarch64/predicates.md (aarch64_plushi_immediate): New.
715 (aarch64_plushi_operand): New.
717 * config/aarch64/aarch64.c (aarch64_expand_compare_and_swap):
718 Force oldval into the rval register for TARGET_LSE; emit the compare
719 during initial expansion so that it may be deleted if unused.
720 (aarch64_gen_atomic_cas): Remove.
721 * config/aarch64/atomics.md (@aarch64_compare_and_swap<SHORT>_lse):
722 Change =&r to +r for operand 0; use match_dup for operand 2;
723 remove is_weak and mod_f operands as unused. Drop the split
725 (@aarch64_atomic_cas<SHORT>): ... this pattern's output; remove.
726 (@aarch64_compare_and_swap<GPI>_lse): Similarly.
727 (@aarch64_atomic_cas<GPI>): Similarly.
729 2018-10-31 Richard Biener <rguenther@suse.de>
731 * tree-eh.c (replace_trapping_overflow): Simplify ABS_EXPR case
734 2018-10-31 Sameera Deshpande <sameera.deshpande@linaro.org>
736 * config/aarch64/aarch64-cores.def (saphira): Use saphira pipeline.
737 * config/aarch64/aarch64.md: Include saphira.md
738 * config/aarch64/saphira.md: New file for pipeline description.
740 2018-10-30 Martin Sebor <msebor@redhat.com>
743 * gimple-ssa-sprintf.c (format_directive): Use %G to include
745 (sprintf_dom_walker::compute_format_length):
746 Avoid setting POSUNDER4K here.
747 (get_destination_size): Handle null argument values.
748 (get_user_idx_format): New function.
749 (sprintf_dom_walker::handle_gimple_call): Handle all printf-like
750 functions, including user-defined with attribute format printf.
751 Use %G to include inlining context.
754 2018-10-30 Jan Hubicka <jh@suse.cz>
756 * params.def (lto-partitions): Bump from 32 to 128.
758 2018-10-30 Jan Hubicka <jh@suse.cz>
761 (free_lang_data_d, add_tree_to_fld_list, fld_worklist_push): Move
763 (free_lang_data_in_type): Forward declare.
764 (fld_type_variant_equal_p): New function.
765 (fld_type_variant): New function
766 (fld_incomplete_types): New hash.
767 (fld_incomplete_type_of): New function
768 (fld_simplfied-type): New function.
769 (free_lang_data_in_decl): Add fld parameter; simplify type of FIELD_DECL
770 (free_lang_data): Allocate and free fld_incomplete_type; update call
771 of free_lang_data_in_decl.
773 2018-10-30 Eric Botcazou <ebotcazou@adacore.com>
775 * gcov.c (output_lines): Remove duplicate line.
777 2018-10-30 Aaron Sawdey <acsawdey@linux.ibm.com>
779 * config/rs6000/rs6000.md (bswapdi2): Force address into register
780 if not in indexed or indirect form.
781 (bswapdi2_load): Change predicate to indexed_or_indirect_operand.
782 (bswapdi2_store): Ditto.
783 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem): New
785 * config/rs6000/rs6000-protos.h (rs6000_force_indexed_or_indirect_mem):
786 Prototype for helper function.
788 2018-10-30 Martin Sebor <msebor@redhat.com>
790 * doc/extend.texi (optimize): Clarify/expand attribute documentation.
791 (target, pragma GCC optimize, pragma GCC target): Ditto.
793 2018-10-30 Jonathan Wakely <jwakely@redhat.com>
795 * doc/extend.texi: Fix prototype and description of
796 __builtin_expect_with_probability.
798 2018-10-30 Michael Ploujnikov <michael.ploujnikov@oracle.com>
800 * cgraph.h (clone_function_name_1): Replaced by new
801 clone_function_name_numbered that takes name as string; for
802 privatize_symbol_name_1 use only.
803 (clone_function_name): Renamed to
804 clone_function_name_numbered to be explicit about numbering.
805 (clone_function_name): New two-argument function that does
806 not number its output.
807 (clone_function_name): New three-argument function that
808 takes a number to append to its output.
809 * cgraphclones.c (duplicate_thunk_for_node):
810 (clone_function_name_1): Renamed.
811 (clone_function_name_numbered): Two new functions.
812 (clone_function_name): Improved documentation.
813 (cgraph_node::create_virtual_clone): Use clone_function_name_numbered.
814 * config/rs6000/rs6000.c (make_resolver_func): Ditto.
815 * final.c (final_scan_insn_1): Use the new clone_function_name
817 * multiple_target.c (create_dispatcher_calls): Ditto.
818 (create_target_clone): Ditto.
819 * omp-expand.c (grid_expand_target_grid_body): Ditto.
820 * omp-low.c (create_omp_child_function_name): Ditto.
821 * omp-simd-clone.c (simd_clone_create): Ditto.
822 * symtab.c (simd_symtab_node::noninterposable_alias): Use the
823 new clone_function_name without numbering.
825 2018-10-30 Richard Earnshaw <rearnsha@arm.com>
827 * alloc-pool.h (base_pool_allocator <TBlockAllocator>::initialize):
828 Assert that the allocation size is not zero.
830 2018-10-30 Richard Biener <rguenther@suse.de>
832 PR tree-optimization/87800
833 * tree-vect-slp.c (vect_build_slp_tree_2): Reject any
834 non-induction or reduction PHIs.
836 2018-10-30 Sameera Deshpande <sameera.deshpande@linaro.org
838 * config/aarch64/falkor-tag-collision-avoidance.c
839 (execute_tag_collision_avoidance): Call df_note_add_problem.
841 2018-10-30 Martin Liska <mliska@suse.cz>
843 * doc/extend.texi: Fix typo in documentation
844 of __builtin_expect_with_probability.
846 2018-10-29 David Malcolm <dmalcolm@redhat.com>
849 * input.c (get_substring_ranges_for_loc): Detect if
850 linemap_resolve_location gives us a NULL map, and reject
853 2018-10-29 Iain Buclaw <ibuclaw@gdcproject.org>
855 * config.gcc (xstormy16-*-elf): Set tm_d_file.
857 2018-10-29 David Malcolm <dmalcolm@redhat.com>
858 Martin Sebor <msebor@redhat.com>
859 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
861 * doc/ux.texi (Quoting): New subsection, adapted from material at
862 https://gcc.gnu.org/wiki/DiagnosticsGuidelines written by
863 MartinSebor and ManuelLopezIbanez.
864 (Fix-it hints): Note that fix-it hints shouldn't be marked for
867 2018-10-29 Kugan Vivekanandarajah <kuganv@linaro.org>
870 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Fix niter
873 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
875 * gcc.target/powerpc/ssse3-check.h: Remove inaccurate comment.
877 2018-10-29 Paul Koning <ni1d@arrl.net>
879 * doc/rtl.texi (CONST_WIDE_INT_ELT): Give correct macro name.
881 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
883 * gcc/config/rs6000/mmintrin.h (_mm_packs_pi16, _mm_packs_pi32,
884 _mm_packs_pu16, _mm_unpackhi_pi8, _mm_unpacklo_pi8, _mm_add_pi8,
885 _mm_add_pi16, _mm_add_pi32, _mm_sub_pi8, _mm_sub_pi16, _mm_sub_pi32,
886 _mm_cmpgt_pi8, _mm_cmpeq_pi16, _mm_cmpgt_pi16, _mm_cmpeq_pi32,
887 _mm_cmpgt_pi32, _mm_adds_pi8, _mm_adds_pi16, _mm_adds_pu8,
888 _mm_adds_pu16, _mm_subs_pi8, _mm_subs_pi16, _mm_subs_pu8,
889 _mm_subs_pu16, _mm_madd_pi16, _mm_mulhi_pi16, _mm_mullo_pi16,
890 _mm_sll_pi16, _mm_sra_pi16, _mm_srl_pi16, _mm_set1_pi16, _mm_set1_pi8):
891 Change 'vector' to '__vector'.
892 * gcc/config/rs6000/xmmintrin.h (_mm_cvtps_pi32, _mm_cvttps_pi32,
893 _mm_cvtps_pi16, _mm_cvtps_pi8, _mm_max_pi16, _mm_max_pu8, _mm_min_pi16,
894 _mm_min_pu8, _mm_mulhi_pu16, _mm_shuffle_pi16, _mm_avg_pu8,
895 _mm_avg_pu16): Likewise. And, whitespace corrections.
897 2018-10-29 Richard Biener <rguenther@suse.de>
899 PR tree-optimization/87785
900 * tree-vect-slp.c (vect_gather_slp_loads): Only gather
903 2018-10-29 Olivier Hainque <hainque@adacore.com>
905 * VXWORKS_NET_LIBS_RTP: New macro, network part of VXWORKS_LIBS_RTP.
907 2018-10-29 Olivier Hainque <hainque@adacore.com>
909 * config/vxworks.h (VXWORKS_PERSONALITY): New VxWorks
910 ports configuration macro, defaults to "gnu".
911 (VXWORKS_OS_CPP_BUILTINS): Use it. Feed TOOL and
912 TOOL_FAMILY instead of _VX_TOOL/_VXTOOL_FAMILY.
914 2018-10-29 Olivier Hainque <hainque@adacore.com>
916 * config/vxworks.h (CLEAR_INSN_CACHE): #undef before #define.
918 2018-10-29 Richard Biener <rguenther@suse.de>
920 PR tree-optimization/87790
921 * tree-vect-slp.c (vect_mark_slp_stmts): Simplify.
922 (vect_make_slp_decision): Adjust.
923 (vect_slp_analyze_bb_1): Likewise.
924 (vect_detect_hybrid_slp_stmts): Properly union SLP type over
927 2018-10-29 Richard Biener <rguenther@suse.de>
929 PR tree-optimization/87785
930 * tree-vect-slp.c (vect_build_slp_tree_2): Remove loads argument
932 (vect_build_slp_tree): Likewise.
933 (vect_gather_slp_loads): New function.
934 (vect_analyze_slp_instance): Gather loads separately from the
937 2018-10-29 Martin Liska <mliska@suse.cz>
939 * Makefile.in: Make dependency to json.o.
940 * doc/gcov.texi: Document new JSON format, remove
941 old intermediate format documentation.
942 * gcov.c (struct function_info): Come up with m_name and
944 (function_info::function_info): Initialize it.
945 (function_info::~function_info): Release it.
946 (main): Rename flag_intermediate_format to flag_json_format.
947 (print_usage): Describe --json-format.
948 (process_args): Set flag_json_format.
949 (output_intermediate_line): Remove.
950 (output_intermediate_json_line): Likewise.
951 (get_gcov_intermediate_filename): Return new extension
953 (output_intermediate_file): Implement JSON emission.
954 (output_json_intermediate_file): Implement JSON emission.
955 (generate_results): Use ::get_name for function name.
956 Handle JSON output file.
957 (read_graph_file): Use ::get_name instead of cplus_demangle.
958 (read_count_file): Likewise.
959 (solve_flow_graph): Likewise.
960 (add_line_counts): Likewise.
961 (accumulate_line_counts): Use new flag_json_format.
962 (output_function_details): Use ::get_name instead of cplus_demangle.
963 (output_lines): Likewise.
964 * json.cc (test_writing_literals): Add new tests.
965 * json.h (class literal): Add new boolean constructor.
967 2018-10-29 Segher Boessenkool <segher@kernel.crashing.org>
969 PR rtl-optimization/87701
970 PR rtl-optimization/87780
971 * combine.c (make_more_copies): Rewrite.
973 2018-10-28 Kugan Vivekanandarajah <kuganv@linaro.org>
975 * doc/generic.texi (ABSU_EXPR): Document.
976 * match.pd (absu(x)*absu(x) -> x*x): Handle.
977 (absu(absu(X)) -> absu(X)): Likewise.
978 (absu(-X) -> absu(X)): Likewise.
979 (absu(X) where X is nonnegative -> X): Likewise.
981 2018-10-28 Iain Buclaw <ibuclaw@gdcproject.org>
983 * Makefile.in (tm_d_file_list, tm_d_include_list): New variables.
984 (TM_D_H, D_TARGET_DEF, D_TARGET_H, D_TARGET_OBJS): New variables.
985 (tm_d.h, cs-tm_d.h, default-d.o): New rules.
986 (d/d-target-hooks-def.h, s-d-target-hooks-def-h): New rules.
987 (s-tm-texi): Also check timestamp on d-target.def.
988 (generated_files): Add TM_D_H and d-target-hooks-def.h.
989 (build/genhooks.o): Also depend on D_TARGET_DEF.
990 * config.gcc (tm_d_file, d_target_objs, target_has_targetdm): New
992 * config/aarch64/aarch64-d.c: New file.
993 * config/aarch64/aarch64-linux.h (GNU_USER_TARGET_D_CRITSEC_SIZE):
995 * config/aarch64/aarch64-protos.h (aarch64_d_target_versions): New
997 * config/aarch64/aarch64.h (TARGET_D_CPU_VERSIONS): Define.
998 * config/aarch64/t-aarch64 (aarch64-d.o): New rule.
999 * config/arm/arm-d.c: New file.
1000 * config/arm/arm-protos.h (arm_d_target_versions): New prototype.
1001 * config/arm/arm.h (TARGET_D_CPU_VERSIONS): Define.
1002 * config/arm/linux-eabi.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
1003 * config/arm/t-arm (arm-d.o): New rule.
1004 * config/default-d.c: New file.
1005 * config/glibc-d.c: New file.
1006 * config/gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
1007 * config/i386/i386-d.c: New file.
1008 * config/i386/i386-protos.h (ix86_d_target_versions): New prototype.
1009 * config/i386/i386.h (TARGET_D_CPU_VERSIONS): Define.
1010 * config/i386/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
1011 (GNU_USER_TARGET_D_CRITSEC_SIZE): Define.
1012 * config/i386/t-i386 (i386-d.o): New rule.
1013 * config/kfreebsd-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
1014 * config/kopensolaris-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
1015 * config/linux-android.h (ANDROID_TARGET_D_OS_VERSIONS): Define.
1016 * config/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
1017 * config/mips/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
1018 * config/mips/mips-d.c: New file.
1019 * config/mips/mips-protos.h (mips_d_target_versions): New prototype.
1020 * config/mips/mips.h (TARGET_D_CPU_VERSIONS): Define.
1021 * config/mips/t-mips (mips-d.o): New rule.
1022 * config/powerpcspe/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
1023 * config/powerpcspe/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
1024 * config/powerpcspe/powerpcspe-d.c: New file.
1025 * config/powerpcspe/powerpcspe-protos.h (rs6000_d_target_versions):
1027 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
1028 Support GNU D by using 0 as the language type.
1029 * config/powerpcspe/powerpcspe.h (TARGET_D_CPU_VERSIONS): Define.
1030 * config/powerpcspe/t-powerpcspe (powerpcspe-d.o): New rule.
1031 * config/riscv/riscv-d.c: New file.
1032 * config/riscv/riscv-protos.h (riscv_d_target_versions): New
1034 * config/riscv/riscv.h (TARGET_D_CPU_VERSIONS): Define.
1035 * config/riscv/t-riscv (riscv-d.o): New rule.
1036 * config/rs6000/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
1037 * config/rs6000/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
1038 * config/rs6000/rs6000-d.c: New file.
1039 * config/rs6000/rs6000-protos.h (rs6000_d_target_versions): New
1041 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
1042 Support GNU D by using 0 as the language type.
1043 * config/rs6000/rs6000.h (TARGET_D_CPU_VERSIONS): Define.
1044 * config/rs6000/t-rs6000 (rs6000-d.o): New rule.
1045 * config/s390/s390-d.c: New file.
1046 * config/s390/s390-protos.h (s390_d_target_versions): New prototype.
1047 * config/s390/s390.h (TARGET_D_CPU_VERSIONS): Define.
1048 * config/s390/t-s390 (s390-d.o): New rule.
1049 * config/sparc/sparc-d.c: New file.
1050 * config/sparc/sparc-protos.h (sparc_d_target_versions): New
1052 * config/sparc/sparc.h (TARGET_D_CPU_VERSIONS): Define.
1053 * config/sparc/t-sparc (sparc-d.o): New rule.
1054 * config/t-glibc (glibc-d.o): New rule.
1055 * configure: Regenerated.
1056 * configure.ac (tm_d_file): New variable.
1057 (tm_d_file_list, tm_d_include_list, d_target_objs): Add substitutes.
1058 * doc/contrib.texi (Contributors): Add self for the D frontend.
1059 * doc/frontends.texi (G++ and GCC): Mention D as a supported language.
1060 * doc/install.texi (Configuration): Mention libphobos as an option for
1061 --enable-shared. Mention d as an option for --enable-languages.
1062 (Testing): Mention check-d as a target.
1063 * doc/invoke.texi (Overall Options): Mention .d, .dd, and .di as file
1064 name suffixes. Mention d as a -x option.
1065 * doc/sourcebuild.texi (Top Level): Mention libphobos.
1066 * doc/standards.texi (Standards): Add section on D language.
1067 * doc/tm.texi: Regenerated.
1068 * doc/tm.texi.in: Add @node for D language and ABI, and @hook for
1069 TARGET_CPU_VERSIONS, TARGET_D_OS_VERSIONS, and TARGET_D_CRITSEC_SIZE.
1070 * dwarf2out.c (is_dlang): New function.
1071 (gen_compile_unit_die): Use DW_LANG_D for D.
1072 (declare_in_namespace): Return module die for D, instead of adding
1073 extra declarations into the namespace.
1074 (gen_namespace_die): Generate DW_TAG_module for D.
1075 (gen_decl_die): Handle CONST_DECLSs for D.
1076 (dwarf2out_decl): Likewise.
1077 (prune_unused_types_walk_local_classes): Handle DW_tag_interface_type.
1078 (prune_unused_types_walk): Handle DW_tag_interface_type same as other
1079 kinds of aggregates.
1080 * gcc.c (default_compilers): Add entries for .d, .dd and .di.
1081 * genhooks.c: Include d/d-target.def.
1083 2018-10-28 Iain Sandoe <iain@sandoe.co.uk>
1086 * config/rs6000/darwin.h (STACK_BOUNDARY): New.
1087 (RS6000_STARTING_FRAME_OFFSET): Adjust to preserve 16byte alignment.
1088 (STACK_DYNAMIC_OFFSET): Likewise.
1090 2018-10-27 Sandra Loosemore <sandra@codesourcery.com>
1093 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Fix
1096 2018-10-26 Tulio Magno Quites Machado Filho <tuliom@linux.ibm.com>
1098 * doc/extend.texi (PowerPC builtins): Fix __builtin_unpack_ibm128
1099 return type and other typos.
1101 2018-10-26 Aaron Sawdey <acsawdey@linux.ibm.com>
1103 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Change to
1104 a shorter sequence with fewer branches.
1105 (emit_final_str_compare_gpr): Ditto.
1107 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
1109 * config/rs6000/tmmintrin.h: New file.
1110 * config.gcc (powerpc*-*-*): Add tmmintrin.h to extra_headers.
1112 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
1114 * config/rs6000/mmintrin.h: Enable 32bit compilation.
1115 * config/rs6000/xmmintrin.h: Likewise.
1117 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
1119 * config/rs6000/xmmintrin.h (_mm_extract_pi16): Fix for big-endian.
1121 2018-10-26 Richard Biener <rguenther@suse.de>
1123 * tree-vect-slp.c (vect_mark_slp_stmts): Add visited hash_set
1125 (vect_mark_slp_stmts_relevant): Likewise.
1126 (vect_detect_hybrid_slp_stmts): Likewise.
1127 (vect_bb_slp_scalar_cost): Likewise.
1128 (vect_remove_slp_scalar_calls): Likewise.
1130 2018-10-26 Jan Hubicka <jh@suse.cz>
1132 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix recursion.
1133 (warn_types_mismatch): Fix walk of DECL_NAME.
1134 (odr_types_equivalent_p): Fix overactive assert.
1136 2018-10-26 Richard Biener <rguenther@suse.de>
1138 PR tree-optimization/87105
1139 * tree-vectorizer.h (_slp_tree::refcnt): New member.
1140 * tree-vect-slp.c (vect_free_slp_tree): Decrement and honor
1142 (vect_create_new_slp_node): Initialize refcnt to one.
1144 (scalar_stmts_set_t, bst_fail): Remove.
1145 (vect_build_slp_tree_2): Add bst_map argument and adjust calls.
1146 (vect_build_slp_tree): Add bst_map argument and lookup
1147 already created SLP nodes.
1148 (vect_print_slp_tree): Handle a SLP graph, print SLP node
1150 (vect_slp_rearrange_stmts): Handle a SLP graph.
1151 (vect_analyze_slp_instance): Adjust and free SLP nodes from
1152 the CSE map. Fix indenting.
1153 (vect_schedule_slp_instance): Add short-cut.
1155 2018-10-26 Martin Liska <mliska@suse.cz>
1158 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Hash with
1159 addr_expr and not with pointers.
1161 2018-10-26 Jan Hubicka <jh@suse.cz>
1163 * tree.c (free_lang_data_in_type): Only check main variants.
1164 * ipa-devirt.c (warn_odr): Make static.
1165 (types_same_for_odr): Drop strict variant.
1166 (types_odr_comparable): Likewise.
1167 (odr_or_derived_type_p): Look for main variants.
1168 (odr_name_hasher::equal): Cleanup comment.
1169 (odr_subtypes_equivalent): Add warn and warned arguments; check main
1171 (type_variants_equivalent_p): break out from ...
1172 (odr_types_equivalent): ... here; go for main variants where needed.
1173 (warn_odr): ... here; turn static.
1174 (warn_types_mismatch): Compare mangled names of main variants.
1175 * ipa-utils.h (types_odr_comparable): Drop strict parameter.
1176 (type_with_linkage_p): Sanity check that we look at main variant.
1177 * lto.c (lto_read_decls): Only consider main variant to be ODR type.
1178 * tree.h (types_same_for_odr): Drop strict argument.
1180 2018-10-26 Richard Biener <rguenther@suse.de>
1182 PR tree-optimization/87746
1183 * tree-vect-data-refs.c (vect_update_misalignment_for_peel):
1184 Simplify and fix WRT strided store groups with size not
1185 equal to step in element count.
1186 (vect_analyze_group_access_1): Dump the whole group.
1188 2018-10-25 Carl Love <cel@us.ibm.com>
1190 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_VSCEDPGT,
1191 P9V_BUILTIN_VEC_VSCEDPLT, P9V_BUILTIN_VEC_VSCEDPEQ,
1192 P9V_BUILTIN_VEC_VSCEDPUO): Rename base overloaded name. Add quad
1193 precicion entry for each overloaded builtin.
1194 * config/rs6000/rs6000-builtin.def (VSCEDPGT, VSCEDPLT, VSCEDPEQ,
1195 VSCEDPUO): Rename overloaded name.
1196 (VSCEDPGT, VSCEQPGT, VSCEDPLT, VSCEQPLT, VSCEDPEQ, VSCEQPEQ,
1197 VSCEDPUO, VSCEQPUO): Add defitions for overloaded builtins.
1198 * config/rs6000/vsx.md (xscmpexpqp_<code>_<mode>): Add
1199 define_expand for xscmpexqp instruction.
1200 (*xscmpexpqp): Add define_insn for the xscmpexqp instruction.
1202 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
1203 Jinsong Ji <jji@us.ibm.com>
1205 * config/rs6000/emmintrin.h (_mm_slli_epi16): Replace deprecated
1206 function with vec_sl.
1207 (_mm_slli_epi32): Likewise.
1208 (_mm_slli_epi64): Likewise.
1209 (_mm_srai_epi16): Replace deprecated function with vec_sra.
1210 (_mm_srai_epi32): Likewise.
1211 (_mm_srli_epi16): Replace deprecated function with vec_sr.
1212 (_mm_srli_epi32): Likewise.
1213 (_mm_srli_epi64): Likewise.
1214 (_mm_sll_epi16): Replace deprecated function with vec_sl.
1215 (_mm_sll_epi32): Likewise.
1216 (_mm_sll_epi64): Likewise.
1217 (_mm_sra_epi16): Replace deprecated function with vec_sra.
1218 (_mm_sra_epi32): Likewise.
1219 (_mm_srl_epi16): Replace deprecated function with vec_sr.
1220 (_mm_srl_epi32): Likewise.
1221 (_mm_srl_epi64): Likewise.
1223 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
1224 Jinsong Ji <jji@us.ibm.com>
1226 * gcc/config/rs6000/emmintrin.h (_mm_sll_epi16): Replace
1227 comparison operators with vec_cmp* for compatibility due to
1228 unfortunate history; clean up formatting and use types more
1230 (_mm_sll_epi32): Likewise.
1231 (_mm_sll_epi64): Likewise.
1232 (_mm_srl_epi16): Likewise.
1233 (_mm_srl_epi32): Likewise.
1234 (_mm_srl_epi64): Likewise.
1236 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
1237 Jinsong Ji <jji@us.ibm.com>
1239 * config/rs6000/emmintrin.h (_mm_sll_epi64): Remove wrong cast.
1240 * config/rs6000/xmmintrin.h (_mm_min_ps): Change m's type to
1241 __vector __bool int. Use vec_cmpgt in preference to deprecated
1242 function vec_vcmpgtfp.
1243 (_mm_max_ps): Likewise.
1245 2018-10-25 Jeff Law <law@redhat.com>
1247 * config/rl78/rl78.c (insn_ok_now): Always re-recognize the insn
1250 2018-10-25 Martin Sebor <msebor@redhat.com>
1252 * doc/extend.texi (aligned): Expand attribute description.
1253 (Alignment): Rename section. Discuss function arguments.
1255 2018-10-25 Jan Hubicka <jh@suse.cz>
1257 * ipa-devirt.c (main_odr_variant): Remove.
1258 (hash_odr_name, types_same_for_odr, types_odr_comparable,
1259 odr_name_hasher::equal, odr_subtypes_equivalent_p):
1260 Drop use of main_odr_variant.
1261 (add_type_duplicate): Silence confused warnings on integer types.
1262 (get_odr_type): Always look for main variant.
1263 (register_odr_type): Simplify.
1265 2018-10-25 Richard Biener <rguenther@suse.de>
1267 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses):
1268 Initialize ng to silence error with release checking bootstrap.
1270 2018-10-25 Richard Biener <rguenther@suse.de>
1272 * tree-if-conv.c: Include tree-ssa-sccvn.h.
1273 (tree_if_conversion): Run CSE on the if-converted loop body.
1275 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
1277 * config/s390/constraints.md (ZL): New constraint.
1278 * config/s390/s390.c (legitimate_pic_operand_p): Accept LARL
1280 * config/s390/s390.md (movdi_larl): Remove.
1281 (movdi_64): Add the LARL alternative.
1283 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
1286 * rtl.c (RTX_CODE_HWINT_P_1): New helper macro.
1287 (RTX_CODE_HWINT_P): New macro.
1288 (rtx_code_size): Use RTX_CODE_HWINT_P ().
1290 2018-10-25 Jan Hubicka <jh@suse.cz>
1292 * ipa-devirt.c (odr_types_equivalent_p): Do not ICE if one of types
1295 2018-10-25 Richard Biener <rguenther@suse.de>
1297 PR tree-optimization/87665
1298 PR tree-optimization/87745
1299 * tree-vectorizer.h (get_earlier_stmt): Remove.
1300 (get_later_stmt): Pick up UID from the original non-pattern stmt.
1302 2018-10-25 Sam Tebbs <sam.tebbs@arm.com>
1304 * options.texi (Deprecated): Move list to Var section.
1306 2018-10-24 Bill Schmidt <wschmidt@linux.ibm.com>
1307 Jinsong Ji <jji@us.ibm.com>
1309 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Change deprecated
1310 __vector long to __vector long long.
1311 (_mm_cvtpd_ps): Likewise.
1312 (_mm_cvttpd_epi32): Likewise.
1313 (_mm_cvtpi32_pd): Likewise.
1314 (_mm_unpackhi_epi64): Likewise.
1315 (_mm_unpacklo_epi64): Likewise.
1317 2018-10-24 Segher Boessenkool <segher@kernel.crashing.org>
1319 PR rtl-optimization/87720
1320 * combine.c (make_more_copies): Skip if the dest is pc_rtx.
1322 2018-10-24 Alexandre Oliva <aoliva@redhat.com>
1324 * gimple-ssa-isolate-paths.c
1325 (find_implicit_erroneous_behavior): Do not change code if the
1326 pass is running for warnings only.
1327 (find_explicit_erroneous_behavior): Likewise.
1329 2018-10-24 Michael Meissner <meissner@linux.ibm.com>
1331 * config/rs6000/rs6000.c (TARGET_MANGLE_DECL_ASSEMBLER_NAME):
1332 Define as rs6000_mangle_decl_assembler_name.
1333 (rs6000_mangle_decl_assembler_name): If the user switched from IBM
1334 long double to IEEE long double, switch the names of the long
1335 double built-in functions to be <func>f128 instead of <func>l.
1337 2018-10-24 Martin Sebor <msebor@redhat.com>
1339 * doc/extend.texi (nonnull): List no-argument form. Reference
1340 -fno-delete-null-pointer-checks and -fisolate-erroneous-paths-attribute.
1342 2018-10-24 Richard Biener <rguenther@suse.de>
1344 * tree-ssa-sccvn.c (do_rpo_vn): Free rpo_state.
1346 2018-10-24 Martin Liska <mliska@suse.cz>
1348 PR tree-optimization/84436
1349 * tree-switch-conversion.c (switch_conversion::contains_same_values_p):
1351 (switch_conversion::contains_linear_function_p): New.
1352 (switch_conversion::build_one_array): Support linear
1353 transformation on input.
1354 * tree-switch-conversion.h (struct switch_conversion): Add
1355 contains_linear_function_p declaration.
1357 2018-10-24 Richard Biener <rguenther@suse.de>
1359 * varasm.c (const_hash_1): Return hash of ADDR_EXPR
1360 if its argument is CONSTANT_CLASS_P.
1362 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
1364 * ipa-utils.h (type_with_linkage_p): No longer check for TYPE_STUB_DECL;
1365 it is wrong for forward declarations.
1367 2018-10-24 Ilya Leoshkevich <iii@linux.ibm.com>
1369 * config/s390/s390.c (s390_check_qrst_address): Add the missing
1370 SYMBOL_REF_P () check.
1372 2018-10-24 Richard Biener <rguenther@suse.de>
1374 PR tree-optimization/87105
1375 * tree-vect-data-refs.c (vect_analyze_group_access_1): Adjust
1376 dump classification.
1377 (vect_analyze_data_ref_accesses): Handle duplicate loads and
1378 stores by splitting the affected group after the fact.
1379 * tree-vect-slp.c (vect_build_slp_tree_2): Dump when we
1380 fail the SLP build because of size constraints.
1382 2018-10-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1384 * configure.ac (gcc_cv_ld_aligned_shf_merge): New test.
1385 * configure: Regenerate.
1386 * config.in: Regenerate.
1387 * varasm.c (mergeable_string_section): Use readonly_data_section
1388 if linker doesn't support SHF_MERGE with alignment > 8.
1389 (mergeable_constant_section): Likewise.
1391 2018-10-24 Richard Biener <rguenther@suse.de>
1393 PR tree-optimization/84013
1394 * tree-ssa-structalias.c (struct msdi_data): New struct for
1395 marshalling data to walk_stmt_load_store_ops.
1396 (maybe_set_dependence_info): Refactor as callback for
1397 walk_stmt_load_store_ops.
1398 (compute_dependence_clique): Set restrict info on all stmt kinds.
1400 2018-10-24 Martin Liska <mliska@suse.cz>
1402 * cgraph.c (cgraph_node::dump):
1403 Remove reduntant dumps and make tp_first_run dump more compact.
1405 2018-10-24 Richard Biener <rguenther@suse.de>
1407 PR tree-optimization/87665
1408 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Adjust
1411 2018-10-12 Jeff Law <law@redhat.com>
1413 * config/h8300/h8300.c (h8300_expand_prologue): Fix stm generation
1416 2018-10-23 Richard Biener <rguenther@suse.de>
1418 * tree-vrp.c (add_assert_info): Guard dump_printf with
1420 * gimple-ssa-evrp-analyze.c
1421 (evrp_range_analyzer::record_ranges_from_incoming_edge):
1422 Use value_range::ignore_equivs_equal_p.
1424 2018-10-23 Richard Biener <rguenther@suse.de>
1426 PR tree-optimization/87105
1427 PR tree-optimization/87608
1428 * passes.def (pass_all_early_optimizations): Add early phi-opt
1430 * tree-ssa-phiopt.c (value_replacement): Ignore NOPs and predicts in
1431 addition to debug stmts.
1432 (tree_ssa_phiopt_worker): Add early_p argument, do only min/max
1433 and abs replacement early.
1434 * tree-cfg.c (gimple_empty_block_p): Likewise.
1436 2018-10-23 Richard Earnshaw <rearnsha@arm.com>
1439 * config.gcc (arm*-*-netbsdelf*): Default to StrongARM if no CPU
1440 specified to configure.
1441 (arm*-*-*): Use ARM7TDMI as the target CPU if no default provided.
1443 2018-10-23 Richard Biener <rguenther@suse.de>
1445 PR tree-optimization/87700
1446 * tree-ssa-copy.c (set_copy_of_val): Fix change detection logic.
1448 2018-10-23 Jakub Jelinek <jakub@redhat.com>
1451 * config/i386/avx512vlintrin.h (_mm_mask_mullo_epi32): Change type of
1452 second argument from __mmask16 to __mmask8.
1453 * config/i386/avx512vlbwintrin.h (_mm_mask_packus_epi32,
1454 _mm_mask_packs_epi32): Likewise.
1455 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i64scatter_ps):
1457 (_mm512_mask_prefetch_i64scatter_pd): Likewise. Formatting fix.
1459 2018-10-23 Richard Biener <rguenther@suse.de>
1461 * tree-vect-stmts.c (vect_analyze_stmt): Fix typo in comment.
1463 2018-10-23 Richard Biener <rguenther@suse.de>
1465 PR tree-optimization/86144
1466 * tree-vect-stmts.c (vect_analyze_stmt): Prefer -mveclibabi
1467 over simd attribute.
1469 2018-10-23 Richard Biener <rguenther@suse.de>
1471 PR tree-optimization/87693
1472 * tree-ssa-threadedge.c (thread_around_empty_blocks): Handle
1473 the case we do not find the taken edge.
1475 2018-10-22 Bill Schmidt <wschmidt@linux.ibm.com>
1476 Jinsong Ji <jji@us.ibm.com>
1478 * config/rs6000/emmintrin.h (_MM_SHUFFLE2): Comment cleanup.
1479 (_mm_store_pd): Use unaligned vector type for pointer cast.
1480 (_mm_maskmoveu_si128): Likewise.
1481 * config/rs6000/xmmintrin.h (__m128_u): New typedef.
1482 (_mm_store_ps): Use unaligned vector type for pointer cast.
1484 2018-10-22 Paul Koning <ni1d@arrl.net>
1486 * symtab.c (symtab_node::increase_alignment): Correct max
1489 2018-10-22 Yury Gribov <tetra2005@gmail.com>
1491 PR tree-optimization/87633
1492 * match.pd: Do not generate unordered integer comparisons.
1494 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
1496 PR rtl-optimization/87600
1497 * combine.c: Add include of expr.h.
1498 (cant_combine_insn_p): Do not combine moves from any hard non-fixed
1499 register to a pseudo.
1500 (make_more_copies): New function, add a copy to a new pseudo after
1501 the moves from hard registers into pseudos.
1502 (rest_of_handle_combine): Declare rebuild_jump_labels_after_combine
1503 later. Call make_more_copies.
1505 2018-10-22 Andrew Stubbs <ams@codesourcery.com>
1507 * lra-constraints.c (process_alt_operands): New local array,
1508 matching_early_clobber. Check matching_early_clobber before
1509 decrementing reject, and set matching_early_clobber after.
1511 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
1514 * config/rs6000/rs6000.c (print_operand_address): For unexpected RTL
1515 call output_addr_const and hope for the best.
1517 2018-10-22 Richard Biener <rguenther@suse.de>
1519 * gimple-ssa-evrp-analyze.c
1520 (evrp_range_analyzer::record_ranges_from_incoming_edge): Be
1521 smarter about what ranges to use.
1522 * tree-vrp.c (add_assert_info): Dump here.
1523 (register_edge_assert_for_2): Instead of here at multiple but
1526 * gcc.dg/tree-ssa/evrp12.c: New testcase.
1527 * gcc.dg/predict-6.c: Adjust.
1528 * gcc.dg/tree-ssa/vrp33.c: Disable EVRP.
1529 * gcc.dg/tree-ssa/vrp02.c: Likewise.
1530 * gcc.dg/tree-ssa/cunroll-9.c: Likewise.
1532 2018-10-22 Steven Bosscher <steven@gcc.gnu.org>
1533 Richard Biener <rguenther@suse.de>
1535 * bitmap.h: Update data structure documentation, including a
1536 description of bitmap views as either linked-lists or splay trees.
1537 (struct bitmap_element_def): Update comments for splay tree bitmaps.
1538 (struct bitmap_head_def): Likewise.
1539 (bitmap_list_view, bitmap_tree_view): New prototypes.
1540 (bitmap_initialize_stat): Initialize a bitmap_head's indx and
1542 (bmp_iter_set_init): Assert the iterated bitmaps are in list form.
1543 (bmp_iter_and_init, bmp_iter_and_compl_init): Likewise.
1544 * bitmap.c (bitmap_elem_to_freelist): Unregister overhead of a
1545 released bitmap element here.
1546 (bitmap_element_free): Remove.
1547 (bitmap_elt_clear_from): Work on splay tree bitmaps.
1548 (bitmap_list_link_element): Renamed from bitmap_element_link. Move
1549 this function similar ones such that linked-list bitmap implementation
1550 functions are grouped.
1551 (bitmap_list_unlink_element): Renamed from bitmap_element_unlink,
1552 and moved for grouping.
1553 (bitmap_list_insert_element_after): Renamed from
1554 bitmap_elt_insert_after, and moved for grouping.
1555 (bitmap_list_find_element): New function spliced from bitmap_find_bit.
1556 (bitmap_tree_link_left, bitmap_tree_link_right,
1557 bitmap_tree_rotate_left, bitmap_tree_rotate_right, bitmap_tree_splay,
1558 bitmap_tree_link_element, bitmap_tree_unlink_element,
1559 bitmap_tree_find_element): New functions for splay-tree bitmap
1561 (bitmap_element_link, bitmap_element_unlink, bitmap_elt_insert_after):
1562 Renamed and moved, see above entries.
1563 (bitmap_tree_listify_from): New function to convert part of a splay
1564 tree bitmap to a linked-list bitmap.
1565 (bitmap_list_view): Convert a splay tree bitmap to linked-list form.
1566 (bitmap_tree_view): Convert a linked-list bitmap to splay tree form.
1567 (bitmap_find_bit): Remove.
1568 (bitmap_clear, bitmap_clear_bit, bitmap_set_bit,
1569 bitmap_single_bit_set_p, bitmap_first_set_bit, bitmap_last_set_bit):
1570 Handle splay tree bitmaps.
1571 (bitmap_copy, bitmap_count_bits, bitmap_and, bitmap_and_into,
1572 bitmap_elt_copy, bitmap_and_compl, bitmap_and_compl_into,
1573 bitmap_compl_and_into, bitmap_elt_ior, bitmap_ior, bitmap_ior_into,
1574 bitmap_xor, bitmap_xor_into, bitmap_equal_p, bitmap_intersect_p,
1575 bitmap_intersect_compl_p, bitmap_ior_and_compl,
1576 bitmap_ior_and_compl_into, bitmap_set_range, bitmap_clear_range,
1577 bitmap_hash): Reject trying to act on splay tree bitmaps. Make
1578 corresponding changes to use linked-list specific bitmap_element
1579 manipulation functions as applicable for efficiency.
1580 (bitmap_tree_to_vec): New function.
1581 (debug_bitmap_elt_file): New function split out from ...
1582 (debug_bitmap_file): ... here. Handle splay tree bitmaps.
1583 (bitmap_print): Likewise.
1585 PR tree-optimization/63155
1586 * tree-ssa-propagate.c (ssa_prop_init): Use tree-view for the
1588 * tree-ssa-coalesce.c (coalesce_ssa_name): Populate used_in_copies
1591 2018-10-22 Martin Liska <mliska@suse.cz>
1593 PR tree-optimization/87686
1595 2018-08-29 Martin Liska <mliska@suse.cz>
1597 * tree-switch-conversion.c (switch_conversion::expand):
1598 Strenghten assumption about gswitch statements.
1600 2018-10-22 Martin Liska <mliska@suse.cz>
1602 * ipa-icf.c (sem_item::compare_attributes): Remove.
1603 (sem_item::compare_referenced_symbol_properties): Use
1604 attribute_list_equal instead.
1605 (sem_function::equals_wpa): Likewise.
1606 * ipa-icf.h: Remove compare_attributes.
1608 2018-10-22 Richard Biener <rguenther@suse.de>
1611 * mem-stats.h (mem_usage::operator==): Fix pasto.
1613 2018-10-22 Richard Biener <rguenther@suse.de>
1615 PR tree-optimization/87640
1616 * tree-vrp.c (set_value_range_with_overflow): Decompose
1618 (extract_range_from_binary_expr_1): Adjust.
1620 2018-10-22 Martin Jambor <mjambor@suse.cz>
1622 * tree-eh.h (stmt_could_throw_p): Add function parameter.
1623 (stmt_can_throw_external): Likewise.
1624 (stmt_can_throw_internal): Likewise.
1625 * tree-eh.c (lower_eh_constructs_2): Pass cfun to stmt_could_throw_p.
1626 (lower_eh_constructs_2): Likewise.
1627 (stmt_could_throw_p): Add fun parameter, use it instead of cfun.
1628 (stmt_can_throw_external): Likewise.
1629 (stmt_can_throw_internal): Likewise.
1630 (maybe_clean_eh_stmt_fn): Pass cfun to stmt_could_throw_p.
1631 (maybe_clean_or_replace_eh_stmt): Pass cfun to stmt_could_throw_p.
1632 (maybe_duplicate_eh_stmt_fn): Pass new_fun to stmt_could_throw_p.
1633 (maybe_duplicate_eh_stmt): Pass cfun to stmt_could_throw_p.
1634 (pass_lower_eh_dispatch::execute): Pass cfun to
1635 stmt_can_throw_external.
1636 (cleanup_empty_eh): Likewise.
1637 (verify_eh_edges): Pass cfun to stmt_could_throw_p.
1638 * cgraph.c (cgraph_edge::set_call_stmt): Pass a function to
1639 stmt_can_throw_external instead of pushing it to cfun.
1640 (symbol_table::create_edge): Likewise.
1641 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Pass cfun to
1642 stmt_can_throw_internal.
1643 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Pass cfun
1644 to stmt_could_throw_p.
1645 * gimple-ssa-store-merging.c (handled_load): Pass cfun to
1646 stmt_can_throw_internal.
1647 (pass_store_merging::execute): Likewise.
1648 * gimple-ssa-strength-reduction.c
1649 (find_candidates_dom_walker::before_dom_children): Pass cfun to
1651 * gimplify-me.c (gimple_regimplify_operands): Pass cfun to
1652 stmt_can_throw_internal.
1653 * ipa-pure-const.c (check_call): Pass cfun to stmt_could_throw_p and
1654 to stmt_can_throw_external.
1655 (check_stmt): Pass cfun to stmt_could_throw_p.
1656 (check_stmt): Pass cfun to stmt_can_throw_external.
1657 (pass_nothrow::execute): Likewise.
1658 * trans-mem.c (expand_call_tm): Pass cfun to stmt_can_throw_internal.
1659 * tree-cfg.c (is_ctrl_altering_stmt): Pass cfun to
1660 stmt_can_throw_internal.
1661 (verify_gimple_in_cfg): Pass cfun to stmt_could_throw_p.
1662 (stmt_can_terminate_bb_p): Pass cfun to stmt_can_throw_external.
1663 (gimple_purge_dead_eh_edges): Pass cfun to stmt_can_throw_internal.
1664 * tree-complex.c (expand_complex_libcall): Pass cfun to
1665 stmt_could_throw_p and to stmt_can_throw_internal.
1666 (expand_complex_multiplication): Pass cfun to stmt_can_throw_internal.
1667 * tree-inline.c (copy_edges_for_bb): Likewise.
1668 (maybe_move_debug_stmts_to_successors): Likewise.
1669 * tree-outof-ssa.c (ssa_is_replaceable_p): Pass cfun to
1671 * tree-parloops.c (oacc_entry_exit_ok_1): Likewise.
1672 * tree-sra.c (scan_function): Pass cfun to stmt_can_throw_external.
1673 * tree-ssa-alias.c (stmt_kills_ref_p): Pass cfun to
1674 stmt_can_throw_internal.
1675 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Likewise.
1676 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Pass cfun to
1678 (mark_aliased_reaching_defs_necessary_1): Pass cfun to
1679 stmt_can_throw_internal.
1680 * tree-ssa-forwprop.c (pass_forwprop::execute): Likewise.
1681 * tree-ssa-loop-im.c (movement_possibility): Pass cfun to
1683 * tree-ssa-loop-ivopts.c (find_givs_in_stmt_scev): Likewise.
1684 (add_autoinc_candidates): Pass cfun to stmt_can_throw_internal.
1685 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
1686 (convert_mult_to_fma_1): Likewise.
1687 (convert_to_divmod): Likewise.
1688 * tree-ssa-phiprop.c (propagate_with_phi): Likewise.
1689 * tree-ssa-pre.c (compute_avail): Pass cfun to stmt_could_throw_p.
1690 * tree-ssa-propagate.c
1691 (substitute_and_fold_dom_walker::before_dom_children): Likewise.
1692 * tree-ssa-reassoc.c (suitable_cond_bb): Likewise.
1693 (maybe_optimize_range_tests): Likewise.
1694 (linearize_expr_tree): Likewise.
1695 (reassociate_bb): Likewise.
1696 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
1697 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Likewise.
1698 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
1699 (handle_char_store): Likewise.
1700 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Pass cfun to
1701 stmt_can_throw_internal.
1702 * tree-vect-patterns.c (check_bool_pattern): Pass cfun to
1704 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
1705 (vectorizable_call): Pass cfun to stmt_can_throw_internal.
1706 (vectorizable_simd_clone_call): Likewise.
1707 * value-prof.c (gimple_ic): Pass cfun to stmt_could_throw_p.
1708 (gimple_stringop_fixed_value): Likewise.
1710 2018-10-22 Ilya Leoshkevich <iii@linux.ibm.com>
1712 * config/s390/s390.c (s390_loadrelative_operand_p): Accept
1713 literal pool references.
1714 (s390_check_qrst_address): Adapt to the new behavior of
1715 s390_loadrelative_operand_p ().
1717 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
1720 * config/i386/sse.md (*andnot<mode>3_bcst): New.
1722 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
1725 * config/i386/sse.md (*<code><mode>3_bcst): New.
1727 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
1730 * config/i386/sse.md (avx512bcst): Updated for V4SI, V2DI, V8SI,
1731 V4DI, V16SI and V8DI.
1732 (*sub<mode>3<mask_name>_bcst): New.
1733 (*add<mode>3<mask_name>_bcst): Likewise.
1735 2018-10-21 Bill Schmidt <wschmidt@linux.ibm.com>
1736 Jinsong Ji <jji@us.ibm.com>
1738 * config/rs6000/emmintrin.h (_mm_movemask_pd): Replace __vector
1739 __m64 with __vector unsigned long long for compatibility.
1740 (_mm_movemask_epi8): Likewise.
1741 * config/rs6000/xmmintrin.h (_mm_cvtps_pi32): Likewise.
1742 (_mm_cvttps_pi32): Likewise.
1743 (_mm_cvtpi32_ps): Likewise.
1744 (_mm_cvtps_pi16): Likewise.
1745 (_mm_loadh_pi): Likewise.
1746 (_mm_storeh_pi): Likewise.
1747 (_mm_movehl_ps): Likewise.
1748 (_mm_movelh_ps): Likewise.
1749 (_mm_loadl_pi): Likewise.
1750 (_mm_storel_pi): Likewise.
1751 (_mm_movemask_ps): Likewise.
1752 (_mm_shuffle_pi16): Likewise.
1754 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1757 * config/i386/avx512fintrin.h (_mm512_fnmsub_round_pd): Use
1758 __builtin_ia32_vfnmsubpd512_mask.
1759 (_mm512_mask_fnmsub_round_pd): Likewise.
1760 (_mm512_fnmsub_pd): Likewise.
1761 (_mm512_mask_fnmsub_pd): Likewise.
1762 (_mm512_maskz_fnmsub_round_pd): Use
1763 __builtin_ia32_vfnmsubpd512_maskz.
1764 (_mm512_maskz_fnmsub_pd): Likewise.
1765 (_mm512_fnmsub_round_ps): Use __builtin_ia32_vfnmsubps512_mask.
1766 (_mm512_mask_fnmsub_round_ps): Likewise.
1767 (_mm512_fnmsub_ps): Likewise.
1768 (_mm512_mask_fnmsub_ps): Likewise.
1769 (_mm512_maskz_fnmsub_round_ps): Use
1770 __builtin_ia32_vfnmsubps512_maskz.
1771 (_mm512_maskz_fnmsub_ps): Likewise.
1772 * config/i386/avx512vlintrin.h (_mm256_mask_fnmsub_pd): Use
1773 __builtin_ia32_vfnmsubpd256_mask.
1774 (_mm256_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256_maskz.
1775 (_mm_mask_fnmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
1776 (_mm_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd128_maskz.
1777 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
1778 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
1779 (_mm256_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_maskz.
1780 (_mm_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_mask.
1781 (_mm_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_maskz.
1782 * config/i386/fmaintrin.h (_mm_fnmsub_pd): Use
1783 __builtin_ia32_vfnmsubpd.
1784 (_mm256_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256.
1785 (_mm_fnmsub_ps): Use __builtin_ia32_vfnmsubps.
1786 (_mm256_fnmsub_ps): Use __builtin_ia32_vfnmsubps256.
1787 (_mm_fnmsub_sd): Use __builtin_ia32_vfnmsubsd3.
1788 (_mm_fnmsub_ss): Use __builtin_ia32_vfnmsubss3.
1789 * config/i386/i386-builtin.def: Add
1790 __builtin_ia32_vfnmsubpd256_mask,
1791 __builtin_ia32_vfnmsubpd256_maskz,
1792 __builtin_ia32_vfnmsubpd128_mask,
1793 __builtin_ia32_vfnmsubpd128_maskz,
1794 __builtin_ia32_vfnmsubps256_mask,
1795 __builtin_ia32_vfnmsubps256_maskz,
1796 __builtin_ia32_vfnmsubps128_mask,
1797 __builtin_ia32_vfnmsubps128_maskz,
1798 __builtin_ia32_vfnmsubpd512_mask,
1799 __builtin_ia32_vfnmsubpd512_maskz,
1800 __builtin_ia32_vfnmsubps512_mask,
1801 __builtin_ia32_vfnmsubps512_maskz, __builtin_ia32_vfnmsubss3,
1802 __builtin_ia32_vfnmsubsd3, __builtin_ia32_vfnmsubps,
1803 __builtin_ia32_vfnmsubpd, __builtin_ia32_vfnmsubps256 and.
1804 __builtin_ia32_vfnmsubpd256.
1805 * config/i386/sse.md (fma4i_fnmsub_<mode>): New.
1806 (<avx512>_fnmsub_<mode>_maskz<round_expand_name>): Likewise.
1807 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_1):
1809 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_2):
1811 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_3):
1813 (fmai_vmfnmsub_<mode><round_name>): Likewise.
1815 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1818 * config/i386/avx512fintrin.h (_mm512_fnmadd_round_pd): Use
1819 __builtin_ia32_vfnmaddpd512_mask.
1820 (_mm512_mask_fnmadd_round_pd): Likewise.
1821 (_mm512_fnmadd_pd): Likewise.
1822 (_mm512_mask_fnmadd_pd): Likewise.
1823 (_mm512_maskz_fnmadd_round_pd): Use
1824 __builtin_ia32_vfnmaddpd512_maskz.
1825 (_mm512_maskz_fnmadd_pd): Likewise.
1826 (_mm512_fnmadd_round_ps): Use __builtin_ia32_vfnmaddps512_mask.
1827 (_mm512_mask_fnmadd_round_ps): Likewise.
1828 (_mm512_fnmadd_ps): Likewise.
1829 (_mm512_mask_fnmadd_ps): Likewise.
1830 (_mm512_maskz_fnmadd_round_ps): Use
1831 __builtin_ia32_vfnmaddps512_maskz.
1832 (_mm512_maskz_fnmadd_ps): Likewise.
1833 * config/i386/avx512vlintrin.h (_mm256_mask_fnmadd_pd): Use
1834 __builtin_ia32_vfnmaddpd256_mask.
1835 (_mm256_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256_maskz.
1836 (_mm_mask_fnmadd_pd): Use __builtin_ia32_vfmaddpd128_mask
1837 (_mm_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd128_maskz.
1838 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
1839 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
1840 (_mm256_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_maskz.
1841 (_mm_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_mask.
1842 (_mm_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_maskz.
1843 * config/i386/fmaintrin.h (_mm_fnmadd_pd): Use
1844 __builtin_ia32_vfnmaddpd.
1845 (_mm256_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256.
1846 (_mm_fnmadd_ps): Use __builtin_ia32_vfnmaddps.
1847 (_mm256_fnmadd_ps): Use __builtin_ia32_vfnmaddps256.
1848 (_mm_fnmadd_sd): Use __builtin_ia32_vfnmaddsd3.
1849 (_mm_fnmadd_ss): Use __builtin_ia32_vfnmaddss3.
1850 * config/i386/i386-builtin.def: Add
1851 __builtin_ia32_vfnmaddpd256_mask,
1852 __builtin_ia32_vfnmaddpd256_maskz,
1853 __builtin_ia32_vfnmaddpd128_mask,
1854 __builtin_ia32_vfnmaddpd128_maskz,
1855 __builtin_ia32_vfnmaddps256_mask,
1856 __builtin_ia32_vfnmaddps256_maskz,
1857 __builtin_ia32_vfnmaddps128_mask,
1858 __builtin_ia32_vfnmaddps128_maskz,
1859 __builtin_ia32_vfnmaddpd512_mask,
1860 __builtin_ia32_vfnmaddpd512_maskz,
1861 __builtin_ia32_vfnmaddps512_mask,
1862 __builtin_ia32_vfnmaddps512_maskz, __builtin_ia32_vfnmaddss3,
1863 __builtin_ia32_vfnmaddsd3, __builtin_ia32_vfnmaddps,
1864 __builtin_ia32_vfnmaddpd, __builtin_ia32_vfnmaddps256 and.
1865 __builtin_ia32_vfnmaddpd256.
1866 * config/i386/sse.md (fma4i_fnmadd_<mode>): New.
1867 (<avx512>_fnmadd_<mode>_maskz<round_expand_name>): Likewise.
1868 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_1):
1870 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_2):
1872 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_3):
1874 (fmai_vmfnmadd_<mode><round_name>): Likewise.
1876 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1879 * config/i386/avx512fintrin.h (_mm512_fmsub_round_pd): Use
1880 __builtin_ia32_vfmsubpd512_mask.
1881 (_mm512_mask_fmsub_round_pd): Likewise.
1882 (_mm512_fmsub_pd): Likewise.
1883 (_mm512_mask_fmsub_pd): Likewise.
1884 (_mm512_maskz_fmsub_round_pd): Use
1885 __builtin_ia32_vfmsubpd512_maskz.
1886 (_mm512_maskz_fmsub_pd): Likewise.
1887 (_mm512_fmsub_round_ps): Use __builtin_ia32_vfmsubps512_mask.
1888 (_mm512_mask_fmsub_round_ps): Likewise.
1889 (_mm512_fmsub_ps): Likewise.
1890 (_mm512_mask_fmsub_ps): Likewise.
1891 (_mm512_maskz_fmsub_round_ps): Use
1892 __builtin_ia32_vfmsubps512_maskz.
1893 (_mm512_maskz_fmsub_ps): Likewise.
1894 * config/i386/avx512vlintrin.h (_mm256_mask_fmsub_pd): Use
1895 __builtin_ia32_vfmsubpd256_mask.
1896 (_mm256_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd256_maskz.
1897 (_mm_mask_fmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
1898 (_mm_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd128_maskz.
1899 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
1900 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
1901 (_mm256_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps256_maskz.
1902 (_mm_mask_fmsub_ps): Use __builtin_ia32_vfmsubps128_mask.
1903 (_mm_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps128_maskz.
1904 * config/i386/fmaintrin.h (_mm_fmsub_pd): Use
1905 __builtin_ia32_vfmsubpd.
1906 (_mm256_fmsub_pd): Use __builtin_ia32_vfmsubpd256.
1907 (_mm_fmsub_ps): Use __builtin_ia32_vfmsubps.
1908 (_mm256_fmsub_ps): Use __builtin_ia32_vfmsubps256.
1909 (_mm_fmsub_sd): Use __builtin_ia32_vfmsubsd3.
1910 (_mm_fmsub_ss): Use __builtin_ia32_vfmsubss3.
1911 * config/i386/i386-builtin.def: Add
1912 __builtin_ia32_vfmsubpd256_mask,
1913 __builtin_ia32_vfmsubpd256_maskz,
1914 __builtin_ia32_vfmsubpd128_mask,
1915 __builtin_ia32_vfmsubpd128_maskz,
1916 __builtin_ia32_vfmsubps256_mask,
1917 __builtin_ia32_vfmsubps256_maskz,
1918 __builtin_ia32_vfmsubps128_mask,
1919 __builtin_ia32_vfmsubps128_maskz,
1920 __builtin_ia32_vfmsubpd512_mask,
1921 __builtin_ia32_vfmsubpd512_maskz,
1922 __builtin_ia32_vfmsubps512_mask,
1923 __builtin_ia32_vfmsubps512_maskz, __builtin_ia32_vfmsubss3,
1924 __builtin_ia32_vfmsubsd3, __builtin_ia32_vfmsubps,
1925 __builtin_ia32_vfmsubpd, __builtin_ia32_vfmsubps256 and.
1926 __builtin_ia32_vfmsubpd256.
1927 * config/i386/sse.md (fma4i_fmsub_<mode>): New.
1928 (<avx512>_fmsub_<mode>_maskz<round_expand_name>): Likewise.
1929 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_1):
1931 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_2):
1933 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_3):
1935 (fmai_vmfmsub_<mode><round_name>): Likewise.
1937 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1939 * config/i386/sse.md (*<plusminus_insn><mode>3<mask_name>_bcst_1):
1940 Remove plus. Renamed to ...
1941 (*sub<mode>3<mask_name>_bcst): This.
1942 (*add<mode>3<mask_name>_bcst_2): Renamede to ...
1943 (*add<mode>3<mask_name>_bcst): This.
1945 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1948 * config/i386/sse.md (*mul<mode>3<mask_name>_bcst): New.
1950 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1953 * i386/avx512vlintrin.h (_mm256_or_epi32): New.
1954 (_mm_or_epi32): Likewise.
1955 (_mm256_xor_epi32): Likewise.
1956 (_mm_xor_epi32): Likewise.
1957 (_mm256_or_epi64): Likewise.
1958 (_mm_or_epi64): Likewise.
1959 (_mm256_xor_epi64): Likewise.
1960 (_mm_xor_epi64): Likewise.
1962 2018-10-20 H.J. Lu <hongjiu.lu@intel.com>
1965 * config/i386/sse.md (*<avx512>_div<mode>3<mask_name>_bcst): New.
1967 2018-10-20 Jakub Jelinek <jakub@redhat.com>
1970 * varasm.c (decode_addr_const): Handle COMPOUND_LITERAL_EXPR.
1972 2018-10-20 Andreas Schwab <schwab@linux-m68k.org>
1974 * doc/ux.texi: Move @section directly after @node.
1976 2018-10-19 Jakub Jelinek <jakub@redhat.com>
1980 * omp-low.c (check_omp_nesting_restrictions): Diagnose ordered without
1981 depend closely nested inside of loop with ordered clause with
1984 2018-10-19 David Malcolm <dmalcolm@redhat.com>
1986 * Makefile.in (TEXI_GCCINT_FILES): Add ux.texi.
1987 * doc/gccint.texi: Include ux.texi and use it in top-level menu.
1988 * doc/ux.texi: New file.
1990 2018-10-19 Segher Boessenkool <segher@kernel.crashing.org>
1992 * config/rs6000/rs6000.h (REG_ALLOC_ORDER): Move 68 (that is, CR0) to
1993 be the first CR field allocated.
1995 2018-10-19 Richard Biener <rguenther@suse.de>
1998 * config/i386/i386.c (ix86_builtin_vectorization_cost): Use
1999 TYPE_VECTOR_SUBPARTS and avoid relying on vector mode.
2001 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
2004 * config/i386/sse.md
2005 (*<plusminus_insn><mode>3<mask_name>_bcst_1): New.
2006 (*add<mode>3<mask_name>_bcst_2): Likewise.
2008 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
2010 * config/i386/sse.md
2011 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
2012 Replace nonimmediate_operand with register_operand.
2013 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
2015 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
2018 2018-10-19 Ilya Leoshkevich <iii@linux.ibm.com>
2020 PR rtl-optimization/87596
2021 * lra-constraints.c (spill_hard_reg_in_range): Use INSN_P () +
2022 lra_get_insn_recog_data () instead of lra_insn_recog_data[]
2023 for instructions in FROM..TO range.
2025 2018-10-19 Eric Botcazou <ebotcazou@adacore.com>
2027 * cfgexpand.c (expand_one_var): Use specific wording in error message
2028 for non-local frame variables.
2029 * stor-layout.c (layout_decl): Do not issue a warning for them.
2031 2018-10-19 Robin Dapp <rdapp@linux.ibm.com>
2033 * haifa-sched.c (priority): Add force_recompute parameter.
2034 (apply_replacement): Call priority () with force_recompute = true.
2035 (restore_pattern): Likewise.
2037 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
2039 * simplify-rtx.c (simplify_subreg): Limit mask of vec_merge to
2040 HOST_BITS_PER_WIDE_INT.
2041 (test_vector_ops_duplicate): Likewise.
2043 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
2046 * config/i386/sse.md (VF_AVX512): New.
2047 (avx512bcst): Likewise.
2048 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
2050 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
2052 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
2055 2018-10-18 Jonathan Wakely <jwakely@redhat.com>
2057 * doc/invoke.texi (-dumpversion): Improve grammar.
2058 (-dumpfullversion): Make more consistent with -dumpversion.
2060 2018-10-18 Uros Bizjak <ubizjak@gmail.com>
2062 * config/i386/i386.c (ix86_emit_fp_unordered_jump):
2063 Set JUMP_LABEL to the jump insn.
2064 (ix86_emit_i387_log1p): Use ix86_expand_branch to expand branch.
2065 Predict emitted jump and add label to jump insn.
2067 2018-10-18 David Malcolm <dmalcolm@redhat.com>
2069 PR tree-optimization/87562
2070 * input.c (get_substring_ranges_for_loc): Use
2071 LRK_SPELLING_LOCATION rather than LRK_MACRO_EXPANSION_POINT when
2072 getting the linemap for the endpoint. Verify that it's either
2073 in the same linemap as the start point's spelling location, or
2074 at least in the same file.
2076 2018-10-18 Richard Biener <rguenther@suse.de>
2078 * config/i386/i386.c (ix86_builtin_vectorization_cost): Do not
2079 feed width-specific load/store costs through ix86_vec_cost.
2080 * config/i386/x86-tune-costs.h (athlon_cost): Adjust.
2081 (k8_cost): Likewise.
2082 (bdver_cost): Likewise.
2083 (znver1_cost): Likewise.
2084 (btver1_cost): Likewise.
2085 (btver2_cost): Likewise.
2087 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
2089 * simplify-rtx.c (simplify_subreg): Call simplify_gen_subreg
2090 to simplify subreg of vec_merge.
2092 2018-10-18 Richard Biener <rguenther@suse.de>
2094 * config/i386/i386.c: Fix costing of vector FMA.
2096 2018-10-18 Richard Biener <rguenther@suse.de>
2098 * config/i386/i386.c (ix86_vec_cost): Remove !parallel path
2100 (ix86_builtin_vectorization_cost): For vec_construct properly
2101 cost insertion into SSE regs.
2102 (...): Adjust calls to ix86_vec_cost.
2104 2018-10-18 Richard Biener <rguenther@suse.de>
2108 2018-02-07 Richard Biener <rguenther@suse.de>
2110 PR tree-optimization/84204
2111 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
2114 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
2117 * simplify-rtx.c (simplify_subreg): Simplify subreg of vec_merge
2119 (test_vector_ops_duplicate): Add test for a scalar subreg of a
2120 VEC_MERGE of a VEC_DUPLICATE.
2122 2018-10-17 Joseph Myers <joseph@codesourcery.com>
2124 * doc/cpp.texi (__STDC_VERSION__): Document C2X handling.
2125 * doc/invoke.texi (-std=c2x, -std=gnu2x): Document new options.
2126 * doc/standards.texi (C Language): Document C2X.
2127 * dwarf2out.c (highest_c_language), config/rl78/rl78.c
2128 (rl78_option_override): Handle "GNU C2X" language name.
2130 2018-10-17 Joseph Myers <joseph@codesourcery.com>
2132 * doc/invoke.texi (-std=c17), doc/standards.texi (C Language):
2133 Document C17 as published in 2018.
2135 2018-10-17 Eric Botcazou <ebotcazou@adacore.com>
2138 * fold-const.c (fold_truth_andor_1): If the right side is not constant,
2139 bail out if both sides do not have the same storage order.
2141 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
2143 * bitmap.c (bitmap_head::dump): New.
2144 * bitmap.h (bitmap_head): Add dump().
2145 * gimple-ssa-evrp-analyze.c
2146 (evrp_range_analyzer::try_find_new_range): Adjust for value_range API.
2147 (evrp_range_analyzer::set_ssa_range_info): Same.
2148 (evrp_range_analyzer::record_ranges_from_phis): Same.
2149 (evrp_range_analyzer::record_ranges_from_stmt): Same.
2150 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Same.
2151 * gimple-ssa-sprintf.c (get_int_range): Same.
2152 (format_integer): Same.
2153 (sprintf_dom_walker::handle_gimple_call): Same.
2154 * ipa-cp.c (ipcp_vr_lattice::meet_with_1): Same.
2155 (ipcp_vr_lattice::top_p): Same.
2156 (ipcp_vr_lattice::bottom_p): Same.
2157 (ipcp_vr_lattice::set_to_bottom): Same.
2158 (ipa_vr_operation_and_type_effects): Same.
2159 (propagate_vr_across_jump_function): Same.
2160 (ipcp_store_vr_results): Same.
2161 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Same.
2162 (ipa_print_node_jump_functions_for_edge): Same.
2163 (ipa_get_value_range): Same.
2164 (ipa_compute_jump_functions_for_edge): Same.
2165 (ipa_write_jump_function): Same.
2166 * tree-ssa-dom.c (simplify_stmt_for_jump_threading): Same.
2167 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
2169 * vr-values.c (set_value_range_to_nonnegative): Same.
2170 (set_value_range_to_truthvalue): Same.
2171 (vr_values::get_value_range): Same.
2172 (vr_values::set_defs_to_varying): Same.
2173 (vr_values::update_value_range): Same.
2174 (symbolic_range_based_on_p): Same.
2175 (vr_values::op_with_boolean_value_range_p): Same.
2176 (vr_values::extract_range_for_var_from_comparison_expr): Same.
2177 (vr_values::extract_range_from_ssa_name): Same.
2178 (vr_values::extract_range_from_binary_expr): Same.
2179 (vr_values::extract_range_from_unary_expr): Same.
2180 (vr_values::extract_range_from_cond_expr): Same.
2181 (vr_values::extract_range_from_comparison): Same.
2182 (vr_values::check_for_binary_op_overflow): Same.
2183 (vr_values::extract_range_basic): Same.
2184 (vr_values::extract_range_from_assignment): Same.
2185 (compare_ranges): Same.
2186 (compare_range_with_value): Same.
2187 (vr_values::adjust_range_with_scev): Same.
2188 (vrp_valueize): Same.
2189 (vrp_valueize_1): Same.
2190 (vr_values::get_vr_for_comparison): Same.
2191 (vr_values::compare_name_with_value): Same.
2192 (vr_values::compare_names): Same.
2193 (vr_values::vrp_evaluate_conditional): Same.
2194 (find_case_label_ranges): Same.
2195 (vr_values::vrp_visit_switch_stmt): Same.
2196 (vr_values::extract_range_from_phi_node): Same.
2197 (vr_values::simplify_div_or_mod_using_ranges): Same.
2198 (vr_values::simplify_bit_ops_using_ranges): Same.
2199 (test_for_singularity): Same.
2200 (range_fits_type_p): Same.
2201 (vr_values::simplify_cond_using_ranges_1): Same.
2202 (vr_values::simplify_switch_using_ranges): Same.
2203 (vr_values::simplify_float_conversion_using_ranges): Same.
2204 (vr_values::two_valued_val_range_p): Same.
2205 (vr_values::add_equivalence): Move to value_range::equiv_add.
2206 * vr-values.h (vr_values::add_equivalence): Remove.
2207 (VR_INITIALIZER): Remove.
2208 * tree-vrp.c (value_range::set): New.
2209 (value_range::equiv_add): New.
2210 (value_range::value_range): New.
2211 (value_range::deep_copy): New.
2212 (value_range::check): New.
2213 (value_range::equal_p): New.
2214 (value_range::ignore_equivs_equal_p): New.
2215 (value_range::operator==): New.
2216 (value_range::operator!=): New.
2217 (value_range::symbolic_p): New.
2218 (value_range::numeric_p): New.
2219 (value_range::set_undefined): New.
2220 (value_range::set_varying): New.
2221 (value_range::may_contain_p): New.
2222 (value_range::equiv_clear): New.
2223 (value_range::singleton_p): New.
2224 (value_range::intersect): New.
2225 (value_range::dump): New.
2226 (value_range::set_and_canonicalize): New.
2227 (set_value_range): Adjust for value_range API.
2228 (set_value_range_to_undefined): Same.
2229 (set_value_range_to_varying): Same.
2230 (set_and_canonicalize_value_range): Same.
2231 (set_value_range_to_nonnull): Same.
2232 (set_value_range_to_null): Same.
2233 (range_is_null): Same.
2234 (range_is_nonnull): Same.
2235 (range_int_cst_p): Same.
2236 (range_int_cst_singleton_p): Same.
2237 (symbolic_range_p): Same.
2238 (range_includes_zero_p): Same.
2239 (value_range_constant_singleton): Same.
2240 (vrp_set_zero_nonzero_bits): Same.
2241 (ranges_from_anti_range): Same.
2242 (extract_range_into_wide_ints): Same.
2243 (extract_range_from_multiplicative_op): Same.
2244 (set_value_range_with_overflow): Same.
2245 (extract_range_from_binary_expr_1): Same.
2246 (extract_range_from_unary_expr): Same.
2247 (dump_value_range): Same.
2248 (debug_value_range): Same.
2249 (vrp_prop::check_array_ref): Same.
2250 (vrp_prop::check_mem_ref): Same.
2251 (vrp_prop::vrp_initialize): Same.
2252 (vrp_prop::visit_stmt): Same.
2253 (intersect_ranges): Same.
2254 (vrp_prop::visit_phi): Same.
2255 (vrp_prop::vrp_finalize): Same.
2256 (determine_value_range_1): Same.
2257 (determine_value_range): Same.
2258 (vrp_intersect_ranges_1): Rename to...
2259 (vrp_intersect_1): this.
2260 (vrp_intersect_ranges): Rename to...
2261 (value_range::intersect_helper): ...this.
2262 (vrp_meet_1): Rename to...
2263 (value_range::union_helper): ...this.
2264 (vrp_meet): Rename to...
2265 (value_range::union_): ...this.
2266 (copy_value_range): Remove.
2267 * tree-vrp.h (struct value_range): Rewrite into a proper class.
2268 (value_range::vrtype): New.
2269 (value_range::type): New.
2270 (value_range::equiv): New.
2271 (value_range::min): New.
2272 (value_range::max): New.
2273 (value_range::varying_p): New.
2274 (value_range::undefined_p): New.
2275 (value_range::null_p): New.
2276 (value_range::equiv_add): New.
2277 (copy_value_range): Remove.
2279 2018-10-17 David Malcolm <dmalcolm@redhat.com>
2281 * Makefile.in (SELFTEST_TARGETS): New.
2282 (selftest) Change from s-selftest-c to $(SELFTEST_TARGETS).
2283 (C_SELFTEST_FLAGS, C_SELFTEST_DEPS, s-selftest-c, selftest-c-gdb)
2284 (selftest-gdb, selftest-c-valgrind, selftest-valgrind): Move to
2286 (CPP_SELFTEST_FLAGS, CPP_SELFTEST_DEPS, s-selftest-c++)
2287 (selftest-c++-gdb, selftest-c++-valgrind): Move to
2289 * configure: Regenerate.
2290 * configure.ac (selftest_languages): New.
2292 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
2294 * tree-vrp.c (extract_range_from_multiplicative_op): Remove
2295 overflow wraps argument.
2296 (extract_range_from_binary_expr_1): Do not pass overflow wraps to
2297 wide_int_range_multiplicative_op.
2298 * wide-int-range.cc (wide_int_range_mult_wrapping): Remove
2299 overflow wraps argument.
2300 (wide_int_range_multiplicative_op): Same.
2301 (wide_int_range_lshift): Same.
2302 (wide_int_range_div): Same.
2303 * wide-int-range.h (wide_int_range_multiplicative_op): Same.
2304 (wide_int_range_lshift): Same.
2305 (wide_int_range_div): Same.
2307 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
2309 * wide-int-range.h (wide_int_range_shift_undefined_p): Adjust to
2310 use sign as argument.
2311 * tree-vrp.c (extract_range_from_binary_expr_1): Pass sign to
2312 wide_int_range_shift_undefined_p.
2314 2018-10-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2316 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>):
2318 (@despeculate_copy<ALLI_TI:mode>): ... This.
2319 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): Remove
2322 2018-10-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2324 * config.gcc: Obsolete *-*-solaris2.10*.
2325 * doc/install.texi (Specific, *-*-solaris2*): Document it.
2327 2018-10-12 Jeff Law <law@redhat.com>
2329 * config/ft32/ft32.md (ft32_general_movsrc_operand): Disable
2330 reg + sym +- const_int addressing modes.
2332 2018-10-15 David Malcolm <dmalcolm@redhat.com>
2334 * common.opt (fdiagnostics-minimum-margin-width=): New option.
2335 * diagnostic-show-locus.c (layout::layout): Apply the minimum
2337 (layout::start_annotation_line): Only print up to 3 of the
2338 margin character, to avoid touching the left-hand side.
2339 (selftest::test_diagnostic_show_locus_fixit_lines): Update for
2340 minimum margin width, as set by test_diagnostic_context's ctor.
2341 (selftest::test_fixit_insert_containing_newline): Likewise.
2342 (selftest::test_fixit_insert_containing_newline_2): Likewise.
2343 (selftest::test_line_numbers_multiline_range): Clear
2344 dc.min_margin_width.
2345 * diagnostic.c (diagnostic_initialize): Initialize
2347 * diagnostic.h (struct diagnostic_context): Add field
2349 * doc/invoke.texi: Add -fdiagnostics-minimum-margin-width=.
2350 * opts.c (common_handle_option): Handle
2351 OPT_fdiagnostics_minimum_margin_width_.
2352 * selftest-diagnostic.c
2353 (selftest::test_diagnostic_context::test_diagnostic_context):
2354 Initialize min_margin_width to 6.
2355 * toplev.c (general_init): Initialize global_dc->min_margin_width.
2357 2018-10-15 David Malcolm <dmalcolm@redhat.com>
2359 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
2360 Fix usage of "error_at_rich_loc" in the comment.
2362 2018-10-15 Renlin Li <renlin.li@arm.com>
2365 * tree-vectorizer.c (try_vectorize_loop_1): Don't use
2366 if-conversioned loop when it contains ifn with types not
2367 supported by backend.
2368 * internal-fn.c (expand_direct_optab_fn): Add an assert.
2369 (direct_internal_fn_supported_p): New helper function.
2370 * internal-fn.h (direct_internal_fn_supported_p): Declare.
2372 2018-10-15 Jakub Jelinek <jakub@redhat.com>
2375 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
2376 Remove OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
2377 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
2379 2018-10-15 Bin Cheng <bin.cheng@linux.alibaba.com>
2381 PR tree-optimization/87022
2382 * tree-loop-distribution.c (pg_add_dependence_edges): Check all
2383 bits in dist vector rather than the first one.
2385 2018-10-15 Richard Biener <rguenther@suse.de>
2388 * tree-ssa-structalias.c (struct vls_data): Add escaped_p member.
2389 (visit_loadstore): When a used restrict tag escaped verify that
2390 the points-to solution of "other" pointers do not include
2392 (compute_dependence_clique): If a used restrict tag escaped
2393 communicated that down to visit_loadstore.
2395 2018-10-15 Andreas Krebbel <krebbel@linux.ibm.com>
2397 * config/s390/s390.c (s390_expand_vec_init): Force vector element
2398 into reg if it isn't a general operand.
2400 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
2403 * config/i386/sse.md (*vec_dupv2di): Add register source to
2406 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
2409 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
2410 Add OPTION_MASK_ISA_AVX512IFMA_UNSET,
2411 OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
2412 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
2414 2018-10-13 Eric Botcazou <ebotcazou@adacore.com>
2416 * dwarf2cfi.c (struct dw_trace_info): Add args_size_defined_for_eh.
2417 (notice_args_size): Set it in the current trace if no insn that can
2418 throw internally has been seen yet.
2419 (connect_traces): When connecting args_size between traces, allow the
2420 incoming values not to match if there is an insn setting it before the
2421 first insn that can throw internally; in that case, force the creation
2422 of a CFI note on this latter insn.
2424 2018-10-13 Jonathan Wakely <jwakely@redhat.com>
2426 * opt-problem.h (opt_wrapper): Use template-argument-list when naming
2427 the base class, because using the injected-class-name was not clearly
2428 specified until DR 176.
2430 2018-10-12 Paul Koning <ni1d@arrl.net>
2432 * config/pdp11/pdp11.md (doloop_end): New expander.
2433 (doloop_end_insn): renamed from "doloop_end".
2434 (addqi3): New pattern.
2435 (subqi3): New pattern.
2436 * config/pdp11/predicates.md (incdec_operand): New predicate.
2438 2018-10-12 Yury Gribov <tetra2005@gmail.com>
2441 * real.c (format_helper::can_represent_integral_type_p): New function
2442 * real.h (format_helper::can_represent_integral_type_p): Ditto.
2443 * match.pd: New pattern.
2445 2018-10-12 Alexandre Oliva <oliva@adacore.com>
2447 * configure.ac: Introduce --enable-large-address-aware
2448 to define MINGW_DEFAULT_LARGE_ADDR_AWARE.
2449 * doc/install.texi: Document it.
2450 * configure, config.in: Rebuilt.
2451 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Define,
2452 based on MINGW_DEFAULT_LARGE_ADDR_AWARE.
2453 (LINK_SPEC): Insert it.
2454 * config/i386/mingw-w64.h: Likewise.
2456 * cygming.h (HAVE_GAS_ALIGNED_COMM): Fallback-define.
2458 2018-10-12 Peter Bergner <bergner@linux.ibm.com>
2460 PR rtl-optimization/87600
2461 * ira-lives (non_conflicting_reg_copy_p): Disable for non LRA targets.
2463 2018-10-12 Paul Koning <ni1d@arrl.net>
2465 * doc/md.texi (doloop_end): Document that the pattern code may
2466 need to check operand mode.
2468 2018-10-12 Wilco Dijkstra <wdijkstr@arm.com>
2470 * config/aarch64/aarch64.md (zero_extendsidi2_aarch64): Add alternatives
2471 to zero-extend between int and floating-point registers.
2472 (load_pair_zero_extendsidi2_aarch64): Add alternative for zero-extended
2473 ldp into floating-point registers. Add type and arch attributes.
2474 (zero_extend<SHORT:mode><GPI:mode>2_aarch64): Add arch attribute.
2475 Use f_loads for type attribute.
2477 2018-10-11 Martin Sebor <msebor@redhat.com>
2479 * doc/extend.texi (attribute packed): Correct typos.
2481 2018-10-11 Martin Sebor <msebor@redhat.com>
2483 * doc/extend.texi (attribute flatten): Mention interaction with
2486 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
2489 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
2490 Set new_decl virtual flag to zero.
2492 2018-10-11 Martin Sebor <msebor@redhat.com>
2495 * doc/extend.texi (attribute format_arg): Discuss using multiple
2496 attributes on a single function.
2498 2018-10-11 Giuliano Belinassi <giuliano.belinassi@usp.br>
2500 PR tree-optimization/86829
2501 * match.pd (sin (atan (x))): New simplification rules.
2502 (cos (atan (x))): Likewise.
2503 * real.c (build_sinatan_real): New function.
2504 * real.h (build_sinatan_real): Prototype.
2506 2018-10-11 Will Schmidt <will_schmidt@vnet.ibm.com>
2508 * config/rs6000/rs6000.c (map_to_integral_tree_type): New helper
2510 (fold_mergeeo_helper): New helper function.
2511 (rs6000_gimple_fold_builtin): Add hooks for vec_mergee and vec_mergeo
2512 intrinsics. Correct some whitespace indentation issues.
2514 2018-10-11 Wilco Dijkstra <wdijkstr@arm.com>
2517 * config/aarch64/aarch64.c (aarch64_mask_and_shift_for_ubfiz_p):
2518 Use HOST_WIDE_INT_1U for shift.
2520 2018-10-11 Doug Rupp <rupp@adacore.com>
2521 Olivier Hainque <hainque@adacore.com>
2523 * config/rs6000/vxworks.h (VXWORKS_RELAX_LINK_SPEC): New macro.
2524 Pass --relax to the linker for RTPs.
2525 (LINK_SPEC): Append VXWORKS_RELAX_LINK_SPEC.
2527 2018-10-11 Andrew Stubbs <ams@codesourcery.com>
2528 Jan Hubicka <jh@suse.cz>
2529 Martin Jambor <mjambor@suse.cz>
2531 * print-rtl.c (print_rtx_operand_codes_E_and_V): Print how many times
2532 the same elements are repeated rather than printing all of them.
2533 * read-rtl.c (rtx_reader::read_rtx_operand): Recognize and expand
2534 "repeated" elements.
2535 * read-rtl-function.c (test_loading_repeat): New function.
2536 (read_rtl_function_c_tests): Call test_loading_repeat.
2537 * rtl-tests.c (test_dumping_repeat): New function.
2538 (rtl_tests_c_tests): Call test_dumping_repeat.
2540 2018-10-11 Richard Biener <rguenther@suse.de>
2542 * config/i386/x86-tune-costs.h (bdver?_memcpy, bdver?_memset,
2543 bdver?_cost): Unify to ...
2544 (bdver_memcpy, bdver_memset, bdver_cost): ... this.
2545 * config/i386/i386.c (processor_cost_table): Adjust.
2547 2018-10-10 Eric Botcazou <ebotcazou@adacore.com>
2550 * cgraphunit.c (cgraph_node::expand_thunk): Force DECL_IGNORED_P on
2551 the thunk when expanding to GIMPLE.
2553 2018-10-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
2555 * varasm.c (mergeable_string_section): Don't try to move zero-length
2556 strings to the merge section.
2558 2018-10-10 Uros Bizjak <ubizjak@gmail.com>
2561 * config/i386/mmx.md (const_vector 0 -> mem splitter): New splitter.
2563 2018-10-10 Jakub Jelinek <jakub@redhat.com>
2566 * config/i386/i386-builtin.def (IX86_BUILTIN_RDPMC): Move from args set
2567 to special_args set.
2569 2018-10-10 Richard Biener <rguenther@suse.de>
2571 * config/i386/sse.md (reduc_plus_scal_v8df, reduc_plus_scal_v4df,
2572 reduc_plus_scal_v2df, reduc_plus_scal_v16sf, reduc_plus_scal_v8sf,
2573 reduc_plus_scal_v4sf): Merge into pattern reducing to half width
2574 and recursing and pattern terminating the recursion on SSE
2575 vector width using ix86_expand_reduc.
2576 (reduc_sminmax_scal_<mode>): Split into part reducing to half
2577 width and recursing and SSE2 vector variant doing the final
2578 reduction with ix86_expand_reduc.
2579 (reduc_uminmax_scal_<mode>): Likewise for the AVX512 variants
2580 with terminating the recursion at AVX level, splitting that
2583 2018-10-09 David Malcolm <dmalcolm@redhat.com>
2585 * genmatch.c (error_cb): Rename to...
2586 (diagnostic_cb): ...this, converting int params to enums.
2587 (fatal_at): Update for renaming.
2588 (warning_at): Likewise.
2590 * input.c (selftest::ebcdic_execution_charset::apply):
2591 Update for renaming of...
2592 (selftest::ebcdic_execution_charset::on_error): ...this, renaming
2594 (selftest::ebcdic_execution_charset::on_diagnostic): ...this,
2595 converting level and reason to enums.
2596 (class selftest::lexer_error_sink): Rename to...
2597 (class selftest::lexer_test_options): ...this, renaming field
2598 "m_errors" to "m_diagnostics".
2599 (selftest::lexer_test_options::apply): Update for renaming of...
2600 (selftest::lexer_test_options::on_error): ...this, renaming to...
2601 (selftest::lexer_test_options::on_diagnostic): ...this
2602 converting level and reason to enums.
2603 (selftest::test_lexer_string_locations_raw_string_unterminated):
2604 Update for renamings.
2605 * opth-gen.awk (struct cpp_reason_option_codes_t): Use enum for
2608 2018-10-09 Paul A. Clarke <pc@us.ibm.com>
2610 * config.gcc (powerpc*-*-*): Add pmmintrin.h to extra_headers.
2611 * config/rs6000/pmmintrin.h: New file.
2613 2018-10-09 Eric Botcazou <ebotcazou@adacore.com>
2615 PR tree-optimization/86659
2616 * gimple-match.h (gimple_match_op constructors): Initialize reverse.
2618 2018-10-09 Richard Biener <rguenther@suse.de>
2620 PR tree-optimization/63155
2621 * tree-ssa-structalias.c: Include tree-ssa.h.
2622 (get_constraint_for_ssa_var): For undefs return nothing_id.
2623 (find_func_aliases): Cleanup PHI handling.
2625 2018-10-09 Richard Biener <rguenther@suse.de>
2627 * tree-vectorizer.c (dump_stmt_cost): Fix cut&paste missing
2630 2018-10-09 Martin Liska <mliska@suse.cz>
2632 * asan.c (asan_emit_stack_protection): If a stack variable
2633 is located in a same file as current function, then emit
2634 line info into variable definition string.
2636 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
2638 * print-rtl.c (rtx_writer::print_rtx_operand_code_i): Print column
2641 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
2643 * cgraphunit.c (cgraph_node::expand_thunk): Do not force DECL_IGNORED_P
2646 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
2648 PR tree-optimization/86659
2649 * gimple-match.h (struct gimple_match_op): Add reverse field.
2650 (gimple_match_op::set_op): New overloaded method.
2651 * gimple-match-head.c (maybe_build_generic_op) <BIT_FIELD_REF>: Set
2652 the REF_REVERSE_STORAGE_ORDER flag on the value.
2653 (gimple_simplify) <GIMPLE_ASSIGN>: For BIT_FIELD_REF, propagate the
2654 REF_REVERSE_STORAGE_ORDER flag and avoid simplifying if it is set.
2656 2018-10-08 Richard Sandiford <richard.sandiford@arm.com>
2659 * gimple-ssa-backprop.c (backprop::intersect_uses): Use
2660 FOR_EACH_IMM_USE_FAST instead of FOR_EACH_IMM_USE_STMT.
2662 2018-10-08 H.J. Lu <hongjiu.lu@intel.com>
2665 * config/i386/avx512fintrin.h (_mm512_mask_fmaddsub_round_pd):
2666 Defined with __builtin_ia32_vfmaddsubpd512_mask.
2668 2018-10-08 Richard Biener <rguenther@suse.de>
2670 * config/i386/x86-tune-costs.h (znver1_cost): Make AVX256 vector loads
2671 cost the same as AVX128 ones.
2673 2018-10-08 Paul Koning <ni1d@arrl.net>
2675 * config/pdp11/pdp11-protos.h (output_block_move): Remove.
2676 (expand_block_move): New function.
2677 * config/pdp11/pdp11.c (output_block_move): Remove.
2678 (expand_block_move): New function.
2679 * config/pdp11/pdp11.h (MOVE_RATIO): New definition.
2680 * config/pdp11/pdp11.md (movmemhi): Use expand_block_move.
2681 (*movmemhi1): Remove.
2683 2018-10-08 Robin Dapp <rdapp@linux.ibm.com>
2685 * config/s390/2827.md: Increase latencies for some FP instructions.
2687 2018-10-08 Richard Biener <rguenther@suse.de>
2689 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
2691 * tree-vectorizer.c (dump_stmt_cost): Add cost param and dump it.
2692 * tree-vectorizer.h (dump_stmt_cost): Adjust.
2693 (add_stmt_cost): Dump return value of the hook.
2695 2018-10-08 Richard Biener <rguenther@suse.de>
2697 PR tree-optimization/63155
2698 * tree-ssa-propagate.c (add_ssa_edge): Do cheap check first.
2699 (ssa_propagation_engine::ssa_propagate): Remove redundant
2700 bitmap bit clearing.
2702 2018-10-05 Peter Bergner <bergner@linux.ibm.com>
2704 PR rtl-optimization/86939
2705 PR rtl-optimization/87479
2706 * ira.h (non_conflicting_reg_copy_p): New prototype.
2707 * ira-lives.c (ignore_reg_for_conflicts): New static variable.
2708 (make_hard_regno_dead): Don't add conflicts for register
2709 ignore_reg_for_conflicts.
2710 (make_object_dead): Likewise.
2711 (non_conflicting_reg_copy_p): New function.
2712 (process_bb_node_lives): Set ignore_reg_for_conflicts for copies.
2713 Remove special conflict handling of REAL_PIC_OFFSET_TABLE_REGNUM.
2714 * lra-lives.c (ignore_reg_for_conflicts): New static variable.
2715 (make_hard_regno_dead): Don't add conflicts for register
2716 ignore_reg_for_conflicts. Remove special conflict handling of
2717 REAL_PIC_OFFSET_TABLE_REGNUM. Remove now unused argument
2718 check_pic_pseudo_p and update callers.
2719 (mark_pseudo_dead): Don't add conflicts for register
2720 ignore_reg_for_conflicts.
2721 (process_bb_lives): Set ignore_reg_for_conflicts for copies.
2723 2018-10-05 Andrew Waterman <andrew@sifive.com>
2724 Jim Wilson <jimw@sifive.com>
2726 * config/riscv/riscv.md (f<quiet_pattern>_quiet<ANYF:mode><X:mode>4):
2727 Add define_expand. Add ! HONOR_SNANS check to current pattern. Add
2728 new pattern using HONOR_SNANS that emits one extra instruction.
2730 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
2732 * config/rs6000/rs6000.md (unnamed mfcr scc_comparison_operator
2733 patterns): Merge SI and DI patterns to a GPR pattern.
2734 (unnamed define_insn and define_split for record form of that): Merge
2735 to a single define_insn_and_split pattern.
2737 2018-10-05 David Malcolm <dmalcolm@redhat.com>
2740 * input.c (expand_location_to_spelling_point): Add param "aspect"
2741 and use rather than hardcoding LOCATION_ASPECT_CARET.
2742 (get_substring_ranges_for_loc): Handle the case of a single token
2743 within a macro expansion.
2744 * input.h (expand_location_to_spelling_point): Add "aspect" param,
2745 defaulting to LOCATION_ASPECT_CARET.
2747 2018-10-05 Paul Koning <ni1d@arrl.net>
2749 * config/pdp11/pdp11.c (TARGET_CXX_GUARD_TYPE): Define.
2750 (TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT): Define.
2751 (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
2752 (pdp11_guard_type): New function.
2754 2018-10-05 Paul Koning <ni1d@arrl.net>
2756 * config/pdp11/pdp11.h (FLOAT_TYPE_SIZE): Always 32.
2757 * config/pdp11/pdp11.opt (mfloat32): Remove.
2759 * doc/invoke.texi (pdp11 -mfloat32): Remove:
2760 (pdp11 -mfloat64): Remove.
2762 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
2764 * config/i386/i386.md (*cmpxf_cc_i387): Remove pattern.
2765 (*cmp<mode>_cc_i387): Ditto.
2766 (*cmpu<mode>_cc_i387): Ditto.
2767 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
2768 * config/i386/i386.c (ix86_expand_fp_compare): Remove
2770 <case IX86_FPCMP_SAHF>: Do not generate pattern with HImode clobber.
2771 Emit x86_sahf_1 pattern.
2772 (ix86_expand_compare): Update call to ix86_expand_fp_compare.
2773 (ix86_expand_carry_flag_compare): Ditto.
2775 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
2777 * config/i386/i386.md (*cmpxf_i387): Change operand 2 predicate
2778 to reg_or_0_operand. Add "C" constraint.
2779 (*cmpxf_cc_i387): Ditto.
2780 (*cmp<mode>_i387): Change operand 2 predicate
2781 to nonimm_or_0_operand. Add "C" constraint.
2782 (*cmp<mode>_cc_i387): Ditto.
2783 (*cmp<mode>_0_i387): Remove insn pattern.
2784 (*cmp<mode>_0_cc_i387): Ditto.
2786 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
2788 * config/i386/constraints.md ("C"): Do not depend on TARGET_SSE.
2789 * config/i386/predicates.md (nonimm_or_0_operand): Rename
2790 from vector_move_operand. Update all uses.
2792 2018-10-05 Martin Sebor <msebor@redhat.com>
2794 PR tree-optimization/87490
2795 * builtins.c (expand_builtin_strnlen): Handle a null data.decl
2798 2018-10-05 Richard Biener <rguenther@suse.de>
2800 PR tree-optimization/63155
2801 * tree-ssa-ccp.c (ccp_propagate::visit_phi): Avoid excess
2802 vertical space in dumpfiles.
2803 * tree-ssa-propagate.h
2804 (ssa_propagation_engine::process_ssa_edge_worklist): Remove.
2805 * tree-ssa-propagate.c (cfg_blocks_back): New global.
2806 (ssa_edge_worklist_back): Likewise.
2807 (curr_order): Likewise.
2808 (cfg_blocks_get): Remove abstraction.
2809 (cfg_blocks_add): Likewise.
2810 (cfg_blocks_empty_p): Likewise.
2811 (add_ssa_edge): Add to current or next worklist based on
2813 (add_control_edge): Likewise.
2814 (ssa_propagation_engine::process_ssa_edge_worklist): Fold
2816 (ssa_propagation_engine::ssa_propagate): ... here. Unify
2817 iteration from CFG and SSA edge worklist so we process
2818 everything in RPO order, prioritizing forward progress
2820 (ssa_prop_init): Allocate new worklists, do not dump
2822 (ssa_prop_fini): Free new worklists.
2824 2018-10-05 Richard Biener <rguenther@suse.de>
2826 * tree-core.h (tree_block::abstract_flag): Remove.
2827 (tree_block::block_num): Make full 32bits.
2828 * tree.def (BLOCK): Remove docs about BLOCK_ABSTRACT.
2829 * tree.h (BLOCK_ABSTRACT): Remove.
2830 * dwarf2out.c (gen_lexical_block_die): Remove dead code
2831 resulting from BLOCK_ABSTRACT being always false.
2832 (gen_inlined_subroutine_die): Likewise.
2833 (gen_block_die): Likewise.
2834 * tree.c (block_ultimate_origin): Likewise.
2835 * tree-pretty-print.c (dump_block_node): Remove code dealing
2836 with BLOCK_ABSTRACT.
2837 * tree-ssa-live.c (dump_scope_block): Likewise.
2838 * tree-streamer-in.c (unpack_ts_block_value_fields): Likewise.
2839 * tree-streamer-out.c (pack_ts_block_value_fields): Likewise.
2841 2018-10-05 Richard Biener <rguenther@suse.de>
2843 * config/i386/i386.c (ix86_add_stmt_cost): When scalar cost
2844 is asked for initialize mode to the component mode of the
2847 2018-10-05 H.J. Lu <hongjiu.lu@intel.com>
2850 * config/i386/gnu-user.h (ASM_SPEC): Don't pass -msse2avx to
2851 assembler for -mavx.
2852 * config/i386/gnu-user64.h (ASM_SPEC): Likewise.
2854 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
2857 * config/rs6000/rs6000-builtin.def (RS6000_BUILTIN_SET_FPSCR_DRN): Use
2859 * config/rs6000/rs6000.md (rs6000_set_fpscr_rn): Require the operand
2860 to be DImode. When using mffscrn, force the operand to a register.
2862 2018-10-04 Uros Bizjak <ubizjak@gmail.com>
2864 * config/i386/i386.md (*fop_<X87MODEF:mode>_2_i387): Macroize insn
2865 from *fop_<MODEF:mode>_2_i387 and *fop_xf_2_i387 using
2866 X87MODEF mode iterator.
2867 (*fop_<X87MODEF:mode>_3_i387): Macroize insn from
2868 *fop_<MODEF:mode>_3_i387 and *fop_xf_3_i387 using
2869 X87MODEF mode iterator.
2871 2018-10-04 Vinay Kumar <vinay.kumar@blackfigtech.com>
2873 * doc/invoke.texi (-Wno-prio-ctor-dtor): Document new warning
2874 -Wno-prio-ctor-dtor.
2876 2018-10-04 David Malcolm <dmalcolm@redhat.com>
2878 * Makefile.in (OBJS): Add opt-problem.o.
2879 * dump-context.h: Include "selftest.h.
2880 (selftest::temp_dump_context): New forward decl.
2881 (class dump_context): Make friend of class
2882 selftest::temp_dump_context.
2883 (dump_context::dump_loc_immediate): New decl.
2884 (class dump_pretty_printer): Move here from dumpfile.c.
2885 (class temp_dump_context): Move to namespace selftest.
2886 (temp_dump_context::temp_dump_context): Add param
2887 "forcibly_enable_dumping".
2888 (selftest::verify_dumped_text):
2889 (ASSERT_DUMPED_TEXT_EQ): Move here from dumpfile.c.
2890 (selftest::verify_item):
2891 (ASSERT_IS_TEXT): Move here from dumpfile.c.
2892 (ASSERT_IS_TREE): Likewise.
2893 (ASSERT_IS_GIMPLE): Likewise.
2894 * dumpfile.c (dump_context::dump_loc): Move immediate dumping
2896 (dump_context::dump_loc_immediate): ...this new function.
2897 (class dump_pretty_printer): Move to dump-context.h.
2898 (dump_switch_p_1): Don't enable MSG_PRIORITY_REEMITTED.
2899 (opt_info_switch_p_1): Enable MSG_PRIORITY_REEMITTED.
2900 (temp_dump_context::temp_dump_context): Move to "selftest"
2901 namespace. Add param "forcibly_enable_dumping", and use it to
2902 conditionalize the use of m_pp;
2903 (selftest::verify_dumped_text): Make non-static.
2904 (ASSERT_DUMPED_TEXT_EQ): Move to dump-context.h.
2905 (selftest::verify_item): Make non-static.
2906 (ASSERT_IS_TEXT): Move to dump-context.h.
2907 (ASSERT_IS_TREE): Likewise.
2908 (ASSERT_IS_GIMPLE): Likewise.
2909 (selftest::test_capture_of_dump_calls): Pass "true" for new
2910 param of temp_dump_context.
2911 * dumpfile.h (enum dump_flag): Add MSG_PRIORITY_REEMITTED, adding
2912 it to MSG_ALL_PRIORITIES. Update values of TDF_COMPARE_DEBUG and
2914 * opt-problem.cc: New file.
2915 * opt-problem.h: New file.
2916 * optinfo-emit-json.cc
2917 (selftest::test_building_json_from_dump_calls): Pass "true" for
2918 new param of temp_dump_context.
2919 * optinfo.cc (optinfo_kind_to_dump_flag): New function.
2920 (optinfo::emit_for_opt_problem): New function.
2921 (optinfo::emit): Clarity which emit_item is used.
2922 * optinfo.h (optinfo::get_dump_location): New accessor.
2923 (optinfo::emit_for_opt_problem): New decl.
2924 (optinfo::emit): Make const.
2925 * selftest-run-tests.c (selftest::run_tests): Call
2926 selftest::opt_problem_cc_tests.
2927 * selftest.h (selftest::opt_problem_cc_tests): New decl.
2928 * tree-data-ref.c (dr_analyze_innermost): Convert return type from
2929 bool to opt_result, converting fprintf messages to
2930 opt_result::failure_at calls. Add "stmt" param for use by the
2932 (create_data_ref): Pass "stmt" to the dr_analyze_innermost call.
2933 (runtime_alias_check_p): Convert return type from bool to
2934 opt_result, converting dump_printf calls to
2935 opt_result::failure_at, using the statement DDR_A for their
2937 (find_data_references_in_stmt): Convert return type from bool to
2938 opt_result, converting "return false" to opt_result::failure_at
2940 * tree-data-ref.h: Include "opt-problem.h".
2941 (dr_analyze_innermost): Convert return type from bool to opt_result,
2942 and add a const gimple * param.
2943 (find_data_references_in_stmt): Convert return type from bool to
2945 (runtime_alias_check_p): Likewise.
2946 * tree-predcom.c (find_looparound_phi): Pass "init_stmt" to
2947 dr_analyze_innermost.
2948 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test):
2949 Convert return type from bool to opt_result, adding a message for
2950 the PARAM_VECT_MAX_VERSION_FOR_ALIAS_CHECKS zero case.
2951 (vect_analyze_data_ref_dependence): Convert return type from bool
2952 to opt_result. Change sense of return type from "false"
2953 effectively meaning "no problems" to "false" meaning a problem,
2954 so that "return false" becomes "return opt_result::success".
2955 Convert "return true" calls to opt_result::failure_at, using
2956 the location of statement A rather than vect_location.
2957 (vect_analyze_data_ref_dependences): Convert return type from bool
2959 (verify_data_ref_alignment): Likewise, converting dump_printf_loc
2960 calls to opt_result::failure_at, using the stmt location rather
2962 (vect_verify_datarefs_alignment): Convert return type from bool
2964 (vect_enhance_data_refs_alignment): Likewise. Split local "stat"
2965 into multiple more-tightly-scoped copies.
2966 (vect_analyze_data_refs_alignment): Convert return type from bool
2968 (vect_analyze_data_ref_accesses): Likewise, converting a
2969 "return false" to a "return opt_result::failure_at", adding a
2971 (vect_prune_runtime_alias_test_list): Convert return type from
2972 bool to opt_result, converting dump_printf_loc to
2973 opt_result::failure_at. Add a %G to show the pertinent statement,
2974 and use the stmt's location rather than vect_location.
2975 (vect_find_stmt_data_reference): Convert return type from
2976 bool to opt_result, converting dump_printf_loc to
2977 opt_result::failure_at, using stmt's location.
2978 (vect_analyze_data_refs): Convert return type from bool to
2979 opt_result. Convert "return false" to "return
2980 opt_result::failure_at", adding messages as needed.
2981 * tree-vect-loop.c (vect_determine_vf_for_stmt_1): Convert return
2982 type from bool to opt_result.
2983 (vect_determine_vf_for_stmt): Likewise.
2984 (vect_determine_vectorization_factor): Likewise, converting
2985 dump_printf_loc to opt_result::failure_at, using location of phi
2986 rather than vect_location.
2987 (vect_analyze_loop_form_1): Convert return type from bool to
2988 opt_result, converting dump_printf_loc calls, retaining the use of
2990 (vect_analyze_loop_form): Convert return type from loop_vec_info
2991 to opt_loop_vec_info.
2992 (vect_analyze_loop_operations): Convert return type from bool to
2993 opt_result, converting dump_printf_loc calls, using the location
2994 of phi/stmt rather than vect_location where available. Convert
2995 various "return false" to "return opt_result::failure_at" with
2996 "unsupported phi" messages.
2997 (vect_get_datarefs_in_loop): Convert return type from bool to
2998 opt_result. Add a message for the
2999 PARAM_LOOP_MAX_DATAREFS_FOR_DATADEPS failure.
3000 (vect_analyze_loop_2): Convert return type from bool to
3001 opt_result. Ensure "ok" is set to a opt_result::failure_at before
3002 each "goto again;", adding new messages where needed.
3003 Add "unsupported grouped {store|load}" messages.
3004 (vect_analyze_loop): Convert return type from loop_vec_info to
3006 * tree-vect-slp.c (vect_analyze_slp): Convert return type from
3008 * tree-vect-stmts.c (process_use): Likewise, converting
3009 dump_printf_loc call and using stmt location, rather than
3011 (vect_mark_stmts_to_be_vectorized): Likeise.
3012 (vect_analyze_stmt): Likewise, adding a %G.
3013 (vect_get_vector_types_for_stmt): Convert return type from bool to
3014 opt_result, converting dump_printf_loc calls and using stmt
3015 location, rather than vect_location.
3016 (vect_get_mask_type_for_stmt): Convert return type from tree to
3017 opt_tree, converting dump_printf_loc calls and using stmt location.
3018 * tree-vectorizer.c: Include "opt-problem.h.
3019 (try_vectorize_loop_1): Flag "Analyzing loop at" dump message as
3020 MSG_PRIORITY_INTERNALS. Convert local "loop_vinfo" from
3021 loop_vec_info to opt_loop_vec_info. If if fails, and dumping is
3022 enabled, use it to report at the top level "couldn't vectorize
3023 loop" followed by the problem.
3024 * tree-vectorizer.h (opt_loop_vec_info): New typedef.
3025 (vect_mark_stmts_to_be_vectorized): Convert return type from bool
3027 (vect_analyze_stmt): Likewise.
3028 (vect_get_vector_types_for_stmt): Likewise.
3029 (tree vect_get_mask_type_for_stmt): Likewise.
3030 (vect_analyze_data_ref_dependences): Likewise.
3031 (vect_enhance_data_refs_alignment): Likewise.
3032 (vect_analyze_data_refs_alignment): Likewise.
3033 (vect_verify_datarefs_alignment): Likewise.
3034 (vect_analyze_data_ref_accesses): Likewise.
3035 (vect_prune_runtime_alias_test_list): Likewise.
3036 (vect_find_stmt_data_reference): Likewise.
3037 (vect_analyze_data_refs): Likewise.
3038 (vect_analyze_loop): Convert return type from loop_vec_info to
3040 (vect_analyze_loop_form): Likewise.
3041 (vect_analyze_slp): Convert return type from bool to opt_result.
3043 2018-10-04 David Malcolm <dmalcolm@redhat.com>
3045 * doc/invoke.texi (-fopt-info): Document new "internals"
3047 * dump-context.h (dump_context::apply_dump_filter_p): New decl.
3048 * dumpfile.c (dump_options): Update for renaming of MSG_ALL to
3050 (optinfo_verbosity_options): Add "internals".
3051 (kind_as_string): Update for renaming of MSG_ALL to MSG_ALL_KINDS.
3052 (dump_context::apply_dump_filter_p): New member function.
3053 (dump_context::dump_loc): Use apply_dump_filter_p rather than
3054 explicitly masking the dump_kind.
3055 (dump_context::begin_scope): Increment the scope depth first. Use
3056 apply_dump_filter_p rather than explicitly masking the dump_kind.
3057 (dump_context::emit_item): Use apply_dump_filter_p rather than
3058 explicitly masking the dump_kind.
3059 (dump_dec): Likewise.
3060 (dump_hex): Likewise.
3061 (dump_switch_p_1): Default to MSG_ALL_PRIORITIES.
3062 (opt_info_switch_p_1): Default to MSG_PRIORITY_USER_FACING.
3063 (opt_info_switch_p): Update handling of default
3064 MSG_OPTIMIZED_LOCATIONS to cope with default of
3065 MSG_PRIORITY_USER_FACING.
3066 (dump_basic_block): Use apply_dump_filter_p rather than explicitly
3067 masking the dump_kind.
3068 (selftest::test_capture_of_dump_calls): Update test_dump_context
3069 instances to use MSG_ALL_KINDS | MSG_PRIORITY_USER_FACING rather
3070 than MSG_ALL. Generalize scope test to be run at all four
3071 combinations of with/without MSG_PRIORITY_USER_FACING and
3072 MSG_PRIORITY_INTERNALS, adding examples of explicit priority
3073 for each of the two values.
3074 * dumpfile.h (enum dump_flag): Add comment about the MSG_* flags.
3075 Rename MSG_ALL to MSG_ALL_KINDS. Add MSG_PRIORITY_USER_FACING,
3076 MSG_PRIORITY_INTERNALS, and MSG_ALL_PRIORITIES, updating the
3077 values for TDF_COMPARE_DEBUG and TDF_ALL_VALUES.
3078 (AUTO_DUMP_SCOPE): Add a note to the comment about the interaction
3079 with MSG_PRIORITY_*.
3080 * tree-vect-loop-manip.c (vect_loop_versioning): Mark versioning
3081 dump messages as MSG_PRIORITY_USER_FACING.
3082 * tree-vectorizer.h (DUMP_VECT_SCOPE): Add a note to the comment
3083 about the interaction with MSG_PRIORITY_*.
3085 2018-10-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3087 * varasm.c (output_constant): Add new parameter merge_strings.
3088 Make strings properly zero terminated in merge string sections.
3089 (mergeable_string_section): Don't fail if the last char is non-zero.
3090 (assemble_variable_contents): Handle merge string sections.
3091 (assemble_variable): Likewise.
3092 (assemble_constant_contents): Likewise.
3093 (output_constant_def_contents): Likewise.
3094 (output_constructor_array_range,
3095 output_constructor_regular_field): Adjust call to output_constant.
3096 (output_object_block): Adjust call to assemble_constant_contents
3097 and assemble_variable_contents.
3099 2018-10-04 Martin Liska <mliska@suse.cz>
3102 * cgraphunit.c (process_function_and_variable_attributes):
3103 Warn about a function with alias attribute and a body.
3105 2018-10-04 Martin Liska <mliska@suse.cz>
3108 * multiple_target.c (redirect_to_specific_clone): New function.
3109 (ipa_target_clone): Use it.
3110 * tree-inline.c: Fix comment.
3112 2018-10-04 David Malcolm <dmalcolm@redhat.com>
3114 * dumpfile.c (gcc::dump_manager::dump_manager): Initialize new
3116 (gcc::dump_manager::~dump_manager): Free m_optinfo_filename.
3117 (gcc::dump_manager::register_pass): New member function, adapted
3118 from loop body in gcc::pass_manager::register_pass, adding a
3119 call to update_dfi_for_opt_info.
3120 (gcc::dump_manager::opt_info_enable_passes): Store the
3121 -fopt-info options into the new fields. Move the loop
3123 (gcc::dump_manager::update_dfi_for_opt_info): ...this new member
3125 * dumpfile.h (struct opt_pass): New forward decl.
3126 (gcc::dump_manager::register_pass): New decl.
3127 (gcc::dump_manager::update_dfi_for_opt_info): New decl.
3128 (class gcc::dump_manager): Add fields "m_optgroup_flags",
3129 "m_optinfo_flags", and "m_optinfo_filename".
3130 * passes.c (gcc::pass_manager::register_pass): Move all of the
3131 dump-handling code to gcc::dump_manager::register_pass.
3133 2018-10-04 Peter Bergner <bergner@linux.ibm.com>
3135 PR rtl-optimization/87466
3136 * target.def (setjmp_preserves_nonvolatile_regs_p): New target hook.
3137 * doc/tm.texi.in (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P): New hook.
3138 * doc/tm.texi: Regenerate.
3139 * ira-lives.c (process_bb_node_lives): Use the new target hook.
3140 * lra-lives.c (process_bb_lives): Likewise.
3141 * config/rs6000/rs6000.c (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P):
3144 2018-10-04 Tamar Christina <tamar.christina@arm.com>
3146 * params.c (add_params): Fix initialization.
3148 2018-10-04 Martin Liska <mliska@suse.cz>
3150 PR gcov-profile/84107
3151 * tree-profile.c (init_ic_make_global_vars):
3152 Remove ic_void_ptr_var and ic_gcov_type_ptr_var.
3153 Come up with new ic_tuple* variables. Emit
3154 __gcov_indirect_call{,_topn} variables.
3155 (gimple_gen_ic_profiler): Access the variable
3157 (gimple_gen_ic_func_profiler): Access
3158 __gcov_indirect_call.callee field.
3159 (gimple_init_gcov_profiler): Use ptr_type_node.
3160 * value-prof.c (gimple_ic): Use ptr_type_node.
3162 2018-10-04 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3164 PR tree-optimization/85787
3165 * ipa-pure-const.c (malloc_candidate_p_1): Move most of malloc_candidate_p
3166 into this function and add support for detecting multiple phis.
3167 (DUMP_AND_RETURN): Move from malloc_candidate_p into top-level macro.
3169 2018-10-04 Martin Liska <mliska@suse.cz>
3172 * ipa-inline.c (inline_to_all_callers_1):
3173 Call ultimate_alias_target for node being inlined.
3175 2018-10-03 Jeff Law <law@redhat.com>
3177 * gimple-ssa-sprintf.c (format_string): Do not hardcode size of
3179 * tree.c (get_typenode_from_name): Moved from fortran/trans-types.c.
3180 * tree.h (get_typenode_from_name): Prototype.
3182 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
3184 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
3185 Change operand 2 predicate to nonimmediate_operand.
3186 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
3188 2018-10-03 Martin Sebor <msebor@redhat.com>
3189 Jeff Law <law@redhat.com>
3191 * gimple-ssa-sprintf.c (struct fmtresult): Add new member and
3193 (get_string_length): Detect unterminated arrays.
3194 (format_string): Same.
3195 (format_directive): Warn about unterminated arrays.
3196 (handle_gimple_call): Mark statements with no_warning as needed.
3198 2018-10-03 Jim Wilson <jimw@sifive.com>
3200 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): For ABI_ILP32E,
3201 also define __riscv_abi_rve. Delete trailing white space.
3203 2018-10-03 Paul Koning <ni1d@arrl.net>
3205 Enable LRA register allocator for PDP11.
3206 * config/pdp11/constraints.md (Q): Use define_memory_constraint.
3209 * config/pdp11/pdp11.c (pdp11_lra_p): New function.
3210 * config/pdp11/pdp11.opt (-mlra): New option.
3211 * doc/invoke.texi (PDP-11 Options): Document -mlra.
3213 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
3215 * config/i386/i386.md (*<absneg:code>extendsfdf2): Remove.
3216 (*<absneg:code>extend<mode>xf2): Ditto.
3218 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
3220 PR tree-optimization/87415
3221 * tree-vrp.c (set_value_range_with_overflow): Special case one bit
3224 2018-10-02 Jeff Law <law@redhat.com>
3226 * gimple-fold.c (get_range_strlen): Only set *nonstr when
3227 an unterminated string is discovered. Bubble up range
3228 even for unterminated strings.
3229 (gimple_fold_builtin_strlen): Do not fold if get_range_strlen
3230 indicates the string was not terminated via NONSTR.
3232 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
3234 * tree-vrp.c (extract_range_from_unary_expr): Special case all
3235 pointer conversions.
3236 Do not do anything special for anti-ranges.
3238 2018-10-03 Jérôme Lambourg <lambourg@adacore.com>
3240 * config/arm/vxworks.h (ARM_TARGET2_DWARF_FORMAT): Adjust to
3241 DW_EH_PE_pcrel | DW_EH_PE_indirect for RTPs.
3243 2018-10-03 Martin Liska <mliska@suse.cz>
3245 PR gcov-profile/86109
3246 * coverage.c (coverage_begin_function): Do not
3247 mark lambdas as artificial.
3248 * tree-core.h (struct GTY): Remove tm_clone_flag
3249 and introduce new lambda_function.
3250 * tree.h (DECL_LAMBDA_FUNCTION): New macro.
3252 2018-10-02 Aaron Sawdey <acsawdey@linux.ibm.com>
3255 * config/rs6000/rs6000-string.c (expand_strn_compare): Check that both
3256 P8_VECTOR and VSX are enabled.
3258 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
3260 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
3261 0x3907 as CPU model number.
3263 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
3265 * common/config/s390/s390-common.c: Rename PF_ARCH12 to PF_Z14.
3266 * config/s390/s390.h (enum processor_flags): Rename PF_ARCH12 to
3267 PF_Z14. Rename TARGET_CPU_ARCH12 to TARGET_CPU_Z14,
3268 TARGET_CPU_ARCH12_P to TARGET_CPU_Z14_P, TARGET_ARCH12 to
3269 TARGET_Z14, and TARGET_ARCH12_P to TARGET_Z14_P.
3270 * config/s390/s390.md: Likewise. Rename also the cpu attribute
3271 value from arch12 to z14.
3273 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
3275 * config/i386/i386.md (fxam<mode>2_i387_with_temp): Remove.
3277 (isinf<mode>2): Ditto.
3279 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
3281 * config/i386/i386.c (ix86_emit_i387_round): Extend op1 to XFmode
3282 before emitting fxam. Perform calculations in XFmode.
3284 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
3286 * match.pd (((X /[ex] A) +- B) * A): New transformation.
3288 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
3291 * fold-const.c (fold_plusminus_mult_expr): Handle complex and vectors.
3292 * tree.c (signed_or_unsigned_type_for): Handle complex.
3294 2018-10-02 Jeff Law <law@redhat.com>
3296 * gimple-fold.c (get_range_strlen): Remove dead code.
3298 2018-10-02 Martin Sebor <msebor@redhat.com>
3299 Jeff Law <law@redhat.com>
3301 * builtins.c (unterminated_array): Add new arguments.
3302 If argument is not terminated, bubble up size and exact
3304 (expand_builtin_strnlen): Detect, avoid expanding
3305 and diagnose unterminated arrays.
3306 (c_strlen): Fill in offset of start of unterminated strings.
3307 * builtins.h (unterminated_array): Update prototype.
3309 2018-10-02 Richard Biener <rguenther@suse.de>
3311 * config/i386/sse.md (reduc_plus_scal_v4df): Avoid the use
3312 of haddv4df, first reduce to SSE width and exploit the fact
3313 that we only need element zero with the reduction result.
3314 (reduc_plus_scal_v2df): Likewise.
3316 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
3318 * dojump.h (do_jump): Delete.
3319 (do_jump_1): Likewise.
3320 (split_comparison): Move around.
3321 * dojump.c (do_jump): Make static.
3322 (do_jump_1): Likewise.
3323 (jumpifnot): Move around.
3324 (jumpifnot_1): Likewise.
3326 (jumpif_1): Likewise.
3327 * expr.c (expand_expr_real_1): Call jumpif[not] instead of do_jump.
3329 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
3331 * reorg.c (make_return_insns): Use emit_copy_of_insn_after for the
3332 insns in the delay slot and add_insn_after for the jump insn.
3334 2018-10-02 Richard Biener <rguenther@suse.de>
3336 * tree-inline.c (expand_call_inline): Use the location of
3337 the callee declaration for the inline-entry marker.
3338 * final.c (notice_source_line): Remove special-casing of
3339 NOTE_INSN_INLINE_ENTRY.
3341 2018-10-01 Carl Love <cel@us.ibm.com>
3344 * config/rs6000/rs6000-builtin.def (__builtin_mffsl): New.
3345 (__builtin_mtfsb0): New.
3346 (__builtin_mtfsb1): New.
3347 ( __builtin_set_fpscr_rn): New.
3348 (__builtin_set_fpscr_drn): New.
3349 * config/rs6000/rs6000.c (rs6000_expand_mtfsb_builtin): Add.
3350 (rs6000_expand_set_fpscr_rn_builtin): Add.
3351 (rs6000_expand_set_fpscr_drn_builtin): Add.
3352 (rs6000_expand_builtin): Add case statement entries for
3353 RS6000_BUILTIN_MTFSB0, RS6000_BUILTIN_MTFSB1,
3354 RS6000_BUILTIN_SET_FPSCR_RN, RS6000_BUILTIN_SET_FPSCR_DRN,
3355 RS6000_BUILTIN_MFFSL.
3356 (rs6000_init_builtins): Add ftype initialization and def_builtin
3357 calls for __builtin_mffsl, __builtin_mtfsb0, __builtin_mtfsb1,
3358 __builtin_set_fpscr_rn, __builtin_set_fpscr_drn.
3359 * config/rs6000.md (rs6000_mtfsb0, rs6000_mtfsb1, rs6000_mffscrn,
3360 rs6000_mffscdrn): Add define_insn.
3361 (rs6000_set_fpscr_rn, rs6000_set_fpscr_drn): Add define_expand.
3362 * doc/extend.texi: Add documentation for the builtins.
3364 2018-10-01 Richard Biener <rguenther@suse.de>
3366 PR tree-optimization/87465
3367 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Fix typo
3368 causing branch miscounts.
3370 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3372 * common/config/aarch64/aarch64-common.c (TARGET_OPTION_DEFAULT_PARAM,
3373 aarch64_option_default_param): New.
3374 (params.h): Include.
3375 (TARGET_OPTION_VALIDATE_PARAM, aarch64_option_validate_param): New.
3376 * config/aarch64/aarch64.c (aarch64_override_options_internal): Simplify
3377 stack-clash protection validation code.
3379 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3381 * params.c (validate_param): New.
3382 (add_params): Use it.
3383 (set_param_value): Refactor param validation into validate_param.
3384 (diagnostic.h): Include.
3385 * diagnostic.h (diagnostic_ready_p): New.
3387 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3389 * params.c (set_param_value):
3390 Add index of parameter being validated.
3391 * common/common-target.def (option_validate_param): New.
3392 * common/common-targhooks.h (default_option_validate_param): New.
3393 * common/common-targhooks.c (default_option_validate_param): New.
3394 * doc/tm.texi.in (TARGET_OPTION_VALIDATE_PARAM): New.
3395 * doc/tm.texi: Regenerate.
3397 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3400 * config/aarch64/aarch64.c (aarch64_override_options_internal):
3401 Add validation for stack-clash parameters and set defaults.
3403 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3406 * configure.ac: Add stack-clash-protection-guard-size.
3407 * doc/install.texi: Document it.
3408 * config.in (DEFAULT_STK_CLASH_GUARD_SIZE): New.
3409 * params.def: Update comment for guard-size.
3410 (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE,
3411 PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Update description.
3412 * configure: Regenerate.
3414 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3417 * config/aarch64/aarch64.h (STACK_CLASH_MIN_BYTES_OUTGOING_ARGS,
3418 STACK_DYNAMIC_OFFSET): New.
3419 * config/aarch64/aarch64.c (aarch64_layout_frame):
3420 Update outgoing args size.
3421 (aarch64_stack_clash_protection_alloca_probe_range,
3422 TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
3424 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3427 * explow.c (anti_adjust_stack_and_probe_stack_clash): Support custom
3429 * target.def (stack_clash_protection_alloca_probe_range): New.
3430 (stack_clash_protection_final_dynamic_probe): Remove.
3431 * targhooks.h (default_stack_clash_protection_alloca_probe_range) New.
3432 (default_stack_clash_protection_final_dynamic_probe): Remove.
3433 * targhooks.c: Likewise.
3434 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
3435 (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE): Remove.
3436 * doc/tm.texi: Regenerate.
3438 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3441 * config/aarch64/aarch64-protos.h (aarch64_output_probe_sve_stack_clash): New.
3442 * config/aarch64/aarch64.c (aarch64_output_probe_sve_stack_clash,
3443 aarch64_clamp_to_uimm12_shift): New.
3444 (aarch64_allocate_and_probe_stack_space): Add SVE specific section.
3445 * config/aarch64/aarch64.md (probe_sve_stack_clash): New.
3447 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3450 * config/aarch64/aarch64.c (aarch64_layout_frame): Add assert.
3452 2018-10-01 Jeff Law <law@redhat.com>
3453 Richard Sandiford <richard.sandiford@linaro.org>
3454 Tamar Christina <tamar.christina@arm.com>
3457 * config/aarch64/aarch64.md
3458 (probe_stack_range): Add k (SP) constraint.
3459 * config/aarch64/aarch64.h (STACK_CLASH_CALLER_GUARD,
3460 STACK_CLASH_MAX_UNROLL_PAGES): New.
3461 * config/aarch64/aarch64.c (aarch64_output_probe_stack_range): Emit
3462 stack probes for stack clash.
3463 (aarch64_allocate_and_probe_stack_space): New.
3464 (aarch64_expand_prologue): Use it.
3465 (aarch64_expand_epilogue): Likewise and update IP regs re-use criteria.
3466 (aarch64_sub_sp): Add emit_move_imm optional param.
3468 2018-10-01 MCC CS <deswurstes@users.noreply.github.com>
3470 PR tree-optimization/87261
3471 * match.pd: Remove trailing whitespace.
3472 Add (x & y) | ~(x | y) -> ~(x ^ y),
3473 (~x | y) ^ (x ^ y) -> x | ~y and (x ^ y) | ~(x | y) -> ~(x & y)
3475 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
3477 * config/arc/arc.md (*add_n): Clean up pattern, update instruction
3479 (ashlsi3_insn): Update instruction constraints.
3480 (ashrsi3_insn): Likewise.
3481 (rotrsi3): Likewise.
3482 (add_shift): Likewise.
3483 * config/arc/constraints.md (Csz): New 32 bit constraint. It
3484 avoids placing in the limm field small constants which, otherwise,
3485 could end into a small instruction.
3487 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
3489 * config/arc/arc.md (maddsidi4_split): Don't use dmac if the
3490 destination register is not odd-even.
3491 (umaddsidi4_split): Likewise.
3493 2018-10-01 Richard Biener <rguenther@suse.de>
3495 * tree-inline.c (expand_call_inline): Store origin of fn
3496 in BLOCK_ABSTRACT_ORIGIN for the inline BLOCK.
3497 * tree.c (block_ultimate_origin): Simplify and do some
3500 2018-09-30 Uros Bizjak <ubizjak@gmail.com>
3502 * config/i386/mmx.md (EMMS): New int iterator.
3503 (emms): New int attribute.
3504 (mmx_<emms>): Macroize insn from *mmx_emms and *mmx_femms using
3505 EMMS int iterator. Explicitly declare clobbers.
3506 (mmx_emms): Remove expander.
3508 * config/i386/predicates.md (emms_operation): Remove predicate.
3509 (vzeroall_pattern): New predicate.
3510 (vzeroupper_pattern): Rename from vzeroupper_operation.
3511 * config/i386/i386.c (ix86_avx_u128_mode_after): Use
3512 vzeroupper_pattern and vzeroall_pattern predicates.
3514 2018-09-30 Peter Bergner <bergner@linux.ibm.com>
3516 PR rtl-optimization/86939
3517 * ira-lives.c (make_hard_regno_born): Rename from this...
3518 (make_hard_regno_live): ... to this. Remove update to conflict
3519 information. Update function comment.
3520 (make_hard_regno_dead): Add conflict information update. Update
3522 (make_object_born): Rename from this...
3523 (make_object_live): ... to this. Remove update to conflict information.
3524 Update function comment.
3525 (make_object_dead): Add conflict information update. Update function
3527 (mark_pseudo_regno_live): Call make_object_live.
3528 (mark_pseudo_regno_subword_live): Likewise.
3529 (mark_hard_reg_dead): Update function comment.
3530 (mark_hard_reg_live): Call make_hard_regno_live.
3531 (process_bb_node_lives): Likewise.
3532 * lra-lives.c (make_hard_regno_born): Rename from this...
3533 (make_hard_regno_live): ... to this. Remove update to conflict
3534 information. Remove now uneeded check_pic_pseudo_p argument.
3535 Update function comment.
3536 (make_hard_regno_dead): Add check_pic_pseudo_p argument and add update
3537 to conflict information. Update function comment.
3538 (mark_pseudo_live): Remove update to conflict information. Update
3540 (mark_pseudo_dead): Add conflict information update.
3541 (mark_regno_live): Call make_hard_regno_live.
3542 (mark_regno_dead): Call make_hard_regno_dead with new arguement.
3543 (process_bb_lives): Call make_hard_regno_live and make_hard_regno_dead.
3545 2018-09-29 H.J. Lu <hongjiu.lu@intel.com>
3548 * config/i386/i386.c (construct_container): Use TImode for
3549 BLKmode values in 2 integer registers.
3551 2018-09-29 Jeff Law <law@redhat.com>
3553 * builtins.c (unterminated_array): Pass in c_strlen_data * to
3554 c_strlen rather than just a tree *.
3555 (c_strlen): Change NONSTR argument to a c_strlen_data pointer.
3556 Update recursive calls appropriately. If caller did not provide a
3557 suitable data pointer, create a local one. When a non-terminated
3558 string is discovered, bubble up information about the string via the
3559 c_strlen_data object.
3560 * builtins.h (c_strlen): Update prototype.
3561 (c_strlen_data): New structure.
3562 * gimple-fold.c (get_range_strlen): Update calls to c_strlen.
3563 For a type 2 call, if c_strlen indicates a non-terminated string
3564 use the length of the non-terminated string.
3565 (gimple_fold_builtin_stpcpy): Update calls to c_strlen.
3567 2018-09-29 Jakub Jelinek <jakub@redhat.com>
3570 * config/i386/avx512fintrin.h (_mm512_abs_pd, _mm512_mask_abs_pd): Use
3571 __m512d type for __A argument rather than __m512.
3573 2018-09-28 John David Anglin <danglin@gcc.gnu.org>
3575 * match.pd (simple_comparison): Don't optimize if either operand is
3576 a function pointer when target needs function pointer canonicalization.
3578 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
3580 * config/rs6000/driver-rs6000.c (asm_names): Adjust the entries for
3581 power5 .. power9 to remove indirection.
3582 * config/rs6000/rs6000.h (ASM_CPU_POWER5_SPEC, ASM_CPU_POWER6_SPEC,
3583 ASM_CPU_POWER7_SPEC, ASM_CPU_POWER8_SPEC, ASM_CPU_POWER9_SPEC,
3584 ASM_CPU_476_SPEC): Delete.
3585 (ASM_CPU_SPEC): Adjust.
3586 (EXTRA_SPECS): Delete asm_cpu_power5, asm_cpu_power6, asm_cpu_power7,
3587 asm_cpu_power8, asm_cpu_power9, asm_cpu_476.
3589 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
3591 * config.in: Delete HAVE_AS_DCI.
3592 * config/powerpcspe/powerpcspe.h: Treat HAVE_AS_DCI as always true.
3593 * config/rs6000/rs6000.h: Ditto.
3594 * configure.ac: Delete HAVE_AS_DCI.
3595 * configure: Regenerate.
3597 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
3599 * config.in (HAVE_AS_LWSYNC): Delete.
3600 * config/powerpcspe/powerpcspe.h (TARGET_LWSYNC_INSTRUCTION): Delete.
3601 * config/powerpcspe/sync.md (*lwsync): Always generate lwsync, never
3603 * config/rs6000/rs6000.h (TARGET_LWSYNC_INSTRUCTION): Delete.
3604 * config/rs6000/sync.md (*lwsync): Always generate lwsync, never do it
3606 * configure.ac: Delete HAVE_AS_LWSYNC.
3607 * configure: Regenerate.
3609 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
3610 Pierre-Marie de Rodat <derodat@adacore.com>
3612 * calls.c (expand_call): Try to do a tail call for thunks at -O0 too.
3613 * cgraph.h (struct cgraph_thunk_info): Add indirect_offset.
3614 (cgraph_node::create_thunk): Add indirect_offset parameter.
3615 (thunk_adjust): Likewise.
3616 * cgraph.c (cgraph_node::create_thunk): Add indirect_offset parameter
3617 and initialize the corresponding field with it.
3618 (cgraph_node::dump): Dump indirect_offset field.
3619 * cgraphclones.c (duplicate_thunk_for_node): Deal with indirect_offset.
3620 * cgraphunit.c (cgraph_node::analyze): Be prepared for external thunks.
3621 (thunk_adjust): Add indirect_offset parameter and deal with it.
3622 (cgraph_node::expand_thunk): Deal with the indirect_offset field and
3623 pass it to thunk_adjust. Do not call the target hook if it's non-zero
3624 or if the thunk is external or local. Fix formatting. Do not chain
3625 the RESULT_DECL to BLOCK_VARS. Pass the static chain to the target,
3626 if any, in the GIMPLE representation.
3627 * ipa-icf.c (sem_function::equals_wpa): Deal with indirect_offset.
3628 * lto-cgraph.c (lto_output_node): Write indirect_offset field.
3629 (input_node): Read indirect_offset field.
3630 * tree-inline.c (expand_call_inline): Pass indirect_offset field in the
3631 call to thunk_adjust.
3632 * tree-nested.c (struct nesting_info): Add thunk_p field.
3633 (create_nesting_tree): Set it.
3634 (convert_all_function_calls): Copy static chain from targets to thunks.
3635 (finalize_nesting_tree_1): Return early for thunks.
3636 (unnest_nesting_tree_1): Do not finalize thunks.
3637 (gimplify_all_functions): Do not gimplify thunks.
3639 2018-09-28 David Malcolm <dmalcolm@redhat.com>
3641 * opt-suggestions.c (option_proposer::build_option_suggestions):
3642 Release "option_values".
3644 2018-09-28 David Malcolm <dmalcolm@redhat.com>
3646 * coverage.c (get_coverage_counts): Convert problem-reporting dump
3647 messages from MSG_OPTIMIZED_LOCATIONS to MSG_MISSED_OPTIMIZATION.
3648 * dumpfile.c (kind_as_string): New function.
3649 (dump_loc): Rather than a hardcoded prefix of "note: ", use
3650 kind_as_string to vary the prefix based on dump_kind.
3651 (selftest::test_capture_of_dump_calls): Update for above.
3653 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
3655 * config/i386/i386.h (SSE_REGNO): Fix check for FIRST_REX_SSE_REG.
3656 (GET_SSE_REGNO): Rename from SSE_REGNO. Update all uses for rename.
3658 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
3660 * config/i386/i386.h (CC_REGNO): Remove FPSR_REG.
3661 * config/i386/i386.c (ix86_fixed_condition_code_regs): Use
3662 INVALID_REGNUM instead of FPSR_REG.
3663 (ix86_md_asm_adjust): Do not clobber FPSR_REG.
3664 * config/i386/i386.md: Update comment of FP compares.
3665 (fldenv): Do not clobber FPSR_REG.
3667 2018-09-28 Richard Biener <rguenther@suse.de>
3669 * tree.h (BLOCK_ORIGIN): New.
3670 * omp-expand.c (grid_expand_target_grid_body): Assign
3671 BLOCK_ORIGIN to BLOCK_ABSTRACT_ORIGIN.
3672 * tree-inline.c (remap_block): Likewise.
3673 * auto-profile.c (get_function_decl_from_block): Simplify
3674 by eliding the BLOCK_ABSTRACT_ORIGIN chasing.
3675 * langhooks.c (lhd_print_error_function): Likewise.
3676 * optinfo-emit-json.cc (optrecord_json_writer::inlining_chain_to):
3678 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
3679 * tree.c (block_nonartificial_location): Likewise.
3680 (block_ultimate_origin): Likewise.
3681 * tree-pretty-print.c (percent_K_format): Likewise. Remove
3682 no longer needed LTO case.
3684 2018-09-28 Andrew Stubbs <ams@codesourcery.com>
3685 Jan Hubicka <jh@suse.cz>
3686 Martin Jambor <mjambor@suse.cz>
3688 * simplify-rtx.c (simplify_merge_mask): New function.
3689 (simplify_ternary_operation): Use it, also see if VEC_MERGEs with the
3690 same masks are used in op1 or op2.
3691 (test_vec_merge): New function.
3692 (test_vector_ops): Call test_vec_merge.
3694 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
3696 * config/sparc/sparc-protos.h (sparc_branch_cost): Declare.
3697 * config/sparc/sparc.h (BRANCH_COST): Call sparc_branch_cost.
3698 * config/sparc/sparc.c (struct processor_costs): Add branch_cost field.
3699 (cypress_costs): Set it.
3700 (supersparc_costs): Likewise.
3701 (hypersparc_costs): Likewise.
3702 (leon_cost): Likewise.
3703 (leon3_costs): Likewise.
3704 (sparclet_costs): Likewise.
3705 (ultrasparc_costs): Likewise.
3706 (ultrasparc_costs): Likewise.
3707 (niagara_costs): Likewise.
3708 (niagara2_costs): Likewise.
3709 (niagara3_costs): Likewise.
3710 (niagara4_costs): Likewise.
3711 (niagara7_costs): Likewise.
3712 (m8_costs): Likewise.
3713 (TARGET_CAN_FOLLOW_JUMP): Define.
3714 (pass_work_around_errata::gate): Minor tweak.
3715 (sparc_option_override): Remove MASK_FSMULD mask for V7 processors.
3716 Do not set both MASK_VIS4 and MASK_VIS4B for M8 processor.
3717 Automaitcally clear MASK_FSMULD mask for V7 processors.
3718 (sparc_can_follow_jump): New static function.
3719 (output_ubranch): Deal with CROSSING_JUMP_P.
3720 (sparc_use_sched_lookahead): Rewrite using switch statement.
3721 (sparc_issue_rate): Reorder.
3722 (sparc_branch_cost): New function.
3724 2018-09-27 Martin Sebor <msebor@redhat.com>
3726 * tree.h (tree_to_shwi): Add attribute nonnull and pure.
3727 (tree_to_poly_int64, tree_to_uhwi, tree_to_poly_uint64): Same.
3728 (int_fits_type_p): Same.
3730 2018-09-27 Uros Bizjak <ubizjak@gmail.com>
3732 * config/i386/i386.md (FPCR_REG): Remove.
3733 (UNSPEC_FLDCW): Remove.
3734 (x86_fnstcw_1): Use (const_int 0) instead of FPCR_REG.
3735 (x86_fldcw_1): Remove insn pattern.
3736 (fnstenv): Do not clobber FPCR_REG.
3738 * config/i386/i386.h (FIXED_REGISTERS) Remove fpsr register.
3739 (CALL_USED_REGISTERS): Ditto.
3740 (REG_ALLOC_ORDER): Ditto.
3741 (REG_CLASS_CONTENTS): Ditto.
3742 (HI_REGISTER_NAMES): Ditto.
3743 (ADDITIONAL_REGISTER_NAMES): Use defines instead
3744 of numerical constants.
3745 * config/i386/i386.c (regclass_map): Remove fpsr register.
3746 (dbx_register_map): Ditto.
3747 (dbx64_register_map): Ditto.
3748 (svr4_dbx_register_map): Ditto.
3749 (print_reg): Do not handle FPCR_REG.
3751 2018-09-27 Segher Boessenkool <segher@kernel.crashing.org>
3754 * config.in (HAVE_AS_CMPB, HAVE_AS_DFP, HAVE_AS_FPRND, HAVE_AS_MFPGPR,
3755 HAVE_AS_POPCNTB, HAVE_AS_POPCNTD, HAVE_AS_POWER8, HAVE_AS_POWER9):
3756 Delete, always treat as true.
3757 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
3758 Ditto. Simplify remaining code.
3759 * config/powerpcspe/powerpcspe.h: Ditto.
3760 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ditto.
3761 Simplify remaining code.
3762 (rs6000_expand_builtin): Ditto.
3763 * config/rs6000/rs6000.h: Ditto.
3764 * configure.ac: Ditto.
3765 * configure: Regenerate.
3767 2018-09-27 Martin Liska <mliska@suse.cz>
3769 * coverage.c (get_coverage_counts): Revert the formatting
3770 of missing profile opt info.
3772 2018-09-27 Richard Biener <rguenther@suse.de>
3776 * dwarf2out.c (set_block_origin_self): Do not mark outermost
3777 block as we do not output that.
3778 (gen_inlined_subroutine_die): Elide the originally outermost
3779 block, matching what we do for concrete instances.
3780 (decls_for_scope): Add parameter specifying whether to recurse
3783 2018-09-27 Andrew Stubbs <ams@codesourcery.com>
3784 Tom de Vries <tom@codesourcery.com>
3788 * expmed.c (emit_cstore): Fix handling of result_mode == BImode and
3789 STORE_FLAG_VALUE == 1.
3791 2018-09-27 Andreas Krebbel <krebbel@linux.ibm.com>
3793 * config/s390/s390.md (PPA_TX_ABORT, PPA_OOO_BARRIER): New
3794 constant definitions.
3795 ("tx_assist"): Replace magic number with PPA_TX_ABORT.
3796 ("*ppa"): Enable pattern also for -march=zEC12 -mno-htm.
3797 ("speculation_barrier"): New expander definition.
3799 2018-09-26 Indu Bhagat <indu.bhagat@oracle.com>
3801 PR gcov-profile/86957
3802 * common.opt: New warning option -Wmissing-profile.
3803 * coverage.c (get_coverage_counts): Add warning for missing .gcda file.
3804 * doc/invoke.texi: Document -Wmissing-profile.
3806 2018-09-26 Jim Wilson <jimw@sifive.com>
3808 * config/riscv/riscv.md (subsi3_extended2): Add J constraint.
3809 (negdi2, negsi2, negsi2_extended, negsi2_extended2): New.
3811 2018-09-26 Martin Sebor <msebor@redhat.com>
3813 * tree.c (zerop): Change return type to bool.
3814 (integer_zerop, integer_onep, integer_each_onep): Same.
3815 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
3816 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
3817 (real_onep, real_minus_onep, chain_index): Same.
3818 (print_type_hash_statistics, type_list_equal): Same.
3819 * tree.h (zerop): Same.
3820 (zerop, integer_zerop, integer_onep, integer_each_onep): Same.
3821 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
3822 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
3823 (real_onep, real_minus_onep, chain_index): Same.
3824 (print_type_hash_statistics, type_list_equal): Same.
3826 2018-09-26 Jim Wilson <jimw@sifive.com>
3828 * config/riscv/riscv.h (FUNCTION_ARG_REGNO_P): Fix comment.
3830 2018-09-26 Jakub Jelinek <jakub@redhat.com>
3833 * config/i386/i386.c: Include debug.h and dwarf2out.h.
3834 (output_indirect_thunk): Emit DW_CFA_def_cfa_offset after the
3837 2018-09-25 Andrew Stubbs <ams@codesourcery.com>
3839 * builtins.c (get_builtin_sync_mem): Force address mode conversion.
3841 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
3843 * config/i386/i386.h (enum reg_class): Remove FP_TOP_SSE_REGS
3844 and FP_SECOND_SSE_REGS.
3845 (REG_CLASS_NAMES): Ditto.
3846 (REG_CLASS_CONTENTS): Ditto.
3847 * config/i386/i386.c (ix86_preferred_reload_class) Do not handle
3848 FP_TOP_SSE_REGS and FP_SECOND_SSE_REGS classes.
3849 (ix86_preferred_output_reload_class): Ditto.
3850 * config/i386/i386.md (fix_trunc<mode>_i387_fisttp): Change "=&1f"
3851 clobber constraint to "=&f".
3852 (fix_truncdi_i387): Ditto.
3853 (lrintxfdi2): Ditto.
3854 (fistdi2_<rounding>): Ditto.
3855 (fpremxf4_i387): Change "=u" constraint to "=f".
3856 (fprem1xf4_i387): Ditto.
3858 (fptanxf4_i387): Ditto.
3859 (fxtractxf3_i387): Ditto.
3860 (fscalexf4_i387): Ditto.
3861 (atan2xf3): Change "u" constraint to "f".
3862 (fyl2xxf3_i387): Ditto.
3863 (fyl2xp1xf3_i387): Ditto.
3865 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
3868 * config/i386/i386.h (NUM_MODES_FOR_MODE_SWITCHING): Update
3869 for removed I387_MASK_PM entity.
3872 2018-09-26 Jeff Law <law@redhat.com>
3874 2018-09-26 Alexey Neyman <stilor@att.net>
3876 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
3877 headers are no longer pulled in by <isl/val.h>.
3879 2018-09-26 Richard Biener <rguenther@suse.de>
3882 * dwarf2out.c (gen_lexical_block_die): Do not equate inline
3883 or concrete instance DIE to the tree. Create abstract origin
3884 attributes also for concrete instances.
3886 2018-09-26 Alexey Neyman <stilor@att.net>
3888 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
3889 headers are no longer pulled in by <isl/val.h>.
3891 2018-09-26 Matthew Malcomson <matthew.malcomson@arm.com>
3893 * config/arm/arm.c (arm_split_compare_and_swap, arm_split_atomic_op):
3894 Use new helper functions.
3895 * config/arm/sync.md (atomic_load<mode>, atomic_store<mode>):
3896 Use new helper functions.
3897 * config/arm/aarch-common-protos.h (aarch_mm_needs_acquire,
3898 aarch_mm_needs_release): New declarations.
3899 * config/arm/aarch-common.c (aarch_mm_needs_acquire,
3900 aarch_mm_needs_release): New.
3902 2018-09-26 Eric Botcazou <ebotcazou@adacore.com>
3904 * config/arm/arm.c (arm_reorg): Skip Thumb reorg pass for thunks.
3905 (arm32_output_mi_thunk): Deal with long calls.
3907 2018-09-26 Richard Biener <rguenther@suse.de>
3911 * tree-inline.c (expand_call_inline): When the location
3912 of the call is UNKNOWN_LOCATION use DECL_SOURCE_LOCATION
3913 or BUILTINS_LOCATION for the BLOCK_SOURCE_LOCATION of
3914 the inserted BLOCK to make inlined_function_outer_scope_p
3916 * dwarf2out.c (add_call_src_coords_attributes): Do not add
3917 coords for reserved locations.
3919 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
3921 * config/rs6000/rs6000.md (*movcc_internal1): Use set_attr_alternative.
3922 (*call_indirect_nonlocal_sysv<mode>): Ditto.
3923 (*call_value_indirect_nonlocal_sysv<mode>): Ditto.
3924 (*sibcall_nonlocal_sysv<mode>): Ditto.
3925 (*sibcall_value_nonlocal_sysv<mode>): Ditto.
3926 (<bd>_<mode>): Ditto.
3927 (<bd>tf_<mode>): Ditto.
3929 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
3931 * config/rs6000/altivec.md (*altivec_mov<mode>): Write the output
3932 control string as a list of templates instead of as C code.
3933 (*altivec_movti): Ditto.
3934 * config/rs6000/darwin.md (movdf_low_di): Ditto.
3936 2018-09-25 Jim Wilson <jimw@sifive.com>
3938 * config/riscv/riscv.c (riscv_split_symbol): Mark auipc label as weak
3939 when target symbol is weak.
3941 2018-09-25 Bernd Edlinger <bernd.edlinger@hotmail.de>
3944 * builtins.c (unterminated_array): Simplify.
3945 * expr.c (string_constant): Handle SSA_NAME. Add more exceptions
3946 where pointer arithmetic is safe.
3948 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
3951 * config/rs6000/altivec.md (altivec_vspltb): Use
3952 const_0_to_15_operand instead of u5bit_cint_operand.
3953 (*altivec_vspltb_internal): Ditto.
3954 (altivec_vspltb_direct): Ditto.
3955 (altivec_vsplth): Use const_0_to_7_operand instead of
3957 (*altivec_vsplth_internal): Ditto.
3958 (altivec_vsplth_direct): Ditto.
3959 (altivec_vspltw): Use const_0_to_3_operand instead of
3961 (*altivec_vspltw_internal): Ditto.
3962 (altivec_vspltw_direct): Ditto.
3963 (altivec_vspltsf): Ditto.
3964 (*altivec_vspltsf_internal): Ditto.
3965 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Handle the
3966 various splats with the proper size immediate. Reorder the various
3967 cases by ascending size of immediate, and put all such together.
3969 2018-09-25 Richard Biener <rguenther@suse.de>
3972 * dwarf2out.c (add_AT_external_die_ref): Remove now redundant
3974 (maybe_create_die_with_external_ref): Do not create
3975 DW_TAG_imported_unit here.
3976 (add_abstract_origin_attribute): Handle external BLOCK refs.
3977 (dwarf2out_abstract_function): Simplify LTO case.
3978 (dwarf2out_early_finish): Create DW_TAG_imported_unit explicitely
3979 rather than using maybe_create_die_with_external_ref.
3981 2018-09-25 Uros Bizjak <ubizjak@gmail.com>
3984 * config/i386/i386.md (frndintxf2_mask_pm): Remove.
3985 (frndintxf2_mask_pm_i387): Ditto.
3986 (nearbyintxf2): Rewrite expander pattern to match rintxf2.
3987 Enable for !flag_trapping_math.
3988 (nearbyint<mode>2): Enable x87 modes for !flag_trapping_math.
3989 Enable SSE modes for TARGET_SSE4_1 and expand them with round insn.
3990 Change operand 1 predicate to nonimmediate_operand.
3991 (attr "i387_cw"): Remove mask_pm.
3992 * config/i386/i386.h (enum ix86_stack_slot): Remove SLOT_CW_MASK_PM.
3993 (enum ix86_entity): Remove I387_MASK_PM.
3994 * config/i386/i386.c (ix86_i387_mode_needed): Do not
3995 handle I387_MASK_PM.
3996 (ix86_mode_needed): Ditto.
3997 (ix86_mode_after): Ditto.
3998 (ix86_mode_entry): Ditto.
3999 (ix86_mode_exit): Ditto.
4000 (emit_i387_cw_initialization): Do not handle I387_CW_MASK_PM.
4002 2018-09-25 Jakub Jelinek <jakub@redhat.com>
4004 * vr-values.c (vr_values::vr_values): Initialize to_remove_edges and
4005 to_update_switch_stmts to vNULL instead of calling create on them
4008 2018-09-25 Richard Biener <rguenther@suse.de>
4010 PR tree-optimization/87402
4011 * tree-ssa-sccvn.c (SSA_VISITED): Remove unused function.
4012 (visit_phi): Re-instantiate handling of supposed to be VARYING
4013 but non-VARYING backedge value.
4015 2018-09-25 Richard Biener <rguenther@suse.de>
4018 * dwarf2out.c (struct sym_off_pair): New.
4019 (external_die_map): New global.
4020 (lookup_decl_die): When in LTO create DIEs lazily from the
4022 (lookup_block_die): New function, create DIEs lazily in LTO.
4023 (equate_block_to_die): New function.
4024 (dwarf2out_die_ref_for_decl): During WPA get the association
4025 from the external DIE map.
4026 (dwarf2out_register_external_die): Record mapping into the
4028 (maybe_create_die_with_external_ref): New function split out from
4029 DIE generation part of old dwarf2out_register_external_die.
4030 (add_abstract_origin_attribute): Do not return the DIE. When
4031 in LTO reference externals directly.
4032 (dwarf2out_abstract_function): When in LTO ignore calls for
4033 decls with external DIEs (already present abstract instances).
4034 (gen_call_site_die): Adjust.
4035 (add_high_low_attributes): Likewise.
4036 (gen_lexical_block_die): Likewise.
4037 (gen_inlined_subroutine_die): Likewie.
4038 (gen_block_die): Likewise.
4039 (dwarf2out_inline_entry): Likewise.
4040 (dwarf2out_early_finish): In LTRANS phase create DW_TAG_imported_unit
4043 2018-09-25 Martin Liska <mliska@suse.cz>
4045 * ipa-fnsummary.c (estimate_node_size_and_time): Scale by two
4046 integers and not by a float value.
4048 2018-09-25 Martin Liska <mliska@suse.cz>
4051 * dbgcnt.c (dbg_cnt_process_single_pair): Return false
4053 * dumpfile.c (dump_enable_all): Remove extra parenthesis.
4054 * gcov-tool.c: Declare the function with ATTRIBUTE_NORETURN.
4055 * godump.c (go_format_type): Remove extra parenthesis.
4057 2018-09-25 Martin Liska <mliska@suse.cz>
4059 * alias.c (set_dest_equal_p): Remove unused function.
4060 * config/i386/i386.c (def_builtin_pure2): Likewise.
4061 * diagnostic-show-locus.c (class layout): Remove
4063 (layout::layout): Likewise here.
4064 * dump-context.h (class temp_dump_context): Likewise.
4065 * dwarf2out.c (add_AT_fde_ref): Remove unused function.
4066 (add_AT_loclistsptr): Likewise.
4067 (add_AT_offset): Likewise.
4068 (get_AT_hi_pc): Likewise.
4069 (is_comdat_die): Likewise.
4070 (type_is_enum): Likewise.
4071 (ceiling): Likewise.
4072 (add_AT_vms_delta): Likewise.
4073 (is_class_die): Likewise.
4074 * edit-context.c (class line_event): Remove unused field.
4075 * graphite-sese-to-poly.c (tree_int_to_gmp): Remove
4077 * ipa-cp.c (ipa_get_vr_lat): Likewise.
4078 * lra-constraints.c (ok_for_index_p_nonstrict): Likewise.
4079 (ok_for_base_p_nonstrict): Likewise.
4080 * tree-chrec.c (is_not_constant_evolution): Likewise.
4081 (chrec_fold_poly_cst): Likewise.
4082 * tree-if-conv.c (has_pred_critical_p): Likewise.
4083 * tree-ssa-coalesce.c (print_exprs): Likewise.
4084 * tree-ssa-pre.c (bitmap_set_contains_expr): Likewise.
4085 * tree-ssa-uninit.c (is_and_or_or_p): Likewise.
4086 * tree-vrp.c (value_ranges_intersect_p): Likewise.
4087 (value_range_nonnegative_p): Likewise.
4089 2018-09-25 Martin Liska <mliska@suse.cz>
4091 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
4092 Do not handle "GNU Pascal".
4093 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
4095 * config/sparc/sparc.c (sparc_pass_by_reference): Remove Pascal
4096 from documentation. Likewise.
4097 * dbxout.c (dbxout_range_type): Likewise.
4098 * doc/cpp.texi: Likewise.
4099 * doc/extend.texi: Likewise.
4100 * doc/frontends.texi: Likewise.
4101 * doc/invoke.texi: Remove Pascal entry.
4102 * tree.def (CLEANUP_POINT_EXPR): Likewise.
4103 * doc/rtl.texi (MODE_FUNCTION): Remove not used entry.
4105 2018-09-25 Martin Liska <mliska@suse.cz>
4108 * doc/invoke.texi: Document all parameters and remove default
4111 2018-09-25 Ilya Leoshkevich <iii@linux.ibm.com>
4114 * rtl.c (rtx_code_size): Take into account that EQ_ATTR_ALT
4115 contains HOST_WIDE_INTs when computing its size.
4117 2018-09-24 Jim Wilson <jimw@sifive.com>
4120 * config/riscv/riscv.h (STACK_BOUNDARY): Test riscv_abi == ABI_ILP32E
4122 (ABI_STACK_BOUNDARY, MAX_ARGS_IN_REGISTERS): Likewise.
4124 2018-09-24 Andrew Pinski <apinski@marvell.com>
4126 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Don't
4127 access prev before checking it for NULLness in the
4128 AARCH64_FUSE_CMP_BRANCH case.
4130 2018-09-24 H.J. Lu <hongjiu.lu@intel.com>
4133 * config/i386/i386.c (rest_of_insert_endbranch): Set
4134 endbr_queued_at_entrance to true and don't insert ENDBR if
4135 x86_function_profiler will be called.
4136 (x86_function_profiler): Insert ENDBR if endbr_queued_at_entrance
4138 * config/i386/i386.h (machine_function): Add
4139 endbr_queued_at_entrance.
4141 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
4143 * genattrtab.c (mk_attr_alt): Use alternative_mask.
4144 (attr_rtx_1): Adjust caching to match the new EQ_ATTR_ALT field
4146 (check_attr_test): Use alternative_mask.
4147 (get_attr_value): Likewise.
4148 (compute_alternative_mask): Use alternative_mask and XWINT.
4149 (make_alternative_compare): Use alternative_mask.
4150 (attr_alt_subset_p): Use XWINT.
4151 (attr_alt_subset_of_compl_p): Likewise.
4152 (attr_alt_intersection): Use alternative_mask and XWINT.
4153 (attr_alt_union): Likewise.
4154 (attr_alt_complement): Use HOST_WIDE_INT and XWINT.
4155 (mk_attr_alt): Use alternative_mask and HOST_WIDE_INT.
4156 (simplify_test_exp): Use alternative_mask and XWINT.
4157 (write_test_expr): Use alternative_mask and XWINT, adjust bit
4158 number calculation to support 64 bits. Generate code that
4159 checks 64-bit masks.
4160 (main): Use alternative_mask.
4161 * rtl.def (EQ_ATTR_ALT): Change field types from ii to ww.
4163 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
4166 * config/s390/s390.c (s390_emit_epilogue): Do not use PARALLEL
4167 RETURN+USE when returning via %r14.
4169 2018-09-24 Martin Liska <mliska@suse.cz>
4171 * gcov.c (output_lines): Print colorization legend
4172 for both flag_use_colors and flag_use_hotness_colors.
4175 2018-09-24 Martin Liska <mliska@suse.cz>
4177 * coverage.c (get_coverage_counts): Use warning_at
4178 with current_function_decl location. Use %qD in warning
4181 2018-09-24 Martin Liska <mliska@suse.cz>
4183 * memory-block.h (memory_block_pool::release): Annotate with
4184 valgrind that the memory is not accessible.
4186 2018-09-24 Martin Liska <mliska@suse.cz>
4189 * asan.c: Make asan_handled_variables extern.
4191 * cfgexpand.c (expand_stack_vars): Make sure
4192 a representative is unpoison if another
4193 variable in the partition is handled by
4194 use-after-scope sanitization.
4196 2018-09-24 Richard Biener <rguenther@suse.de>
4198 PR tree-optimization/63155
4199 * tree-ssa-propagate.c (add_ssa_edge): Avoid adding PHIs to
4200 the worklist when the edge of the respective argument isn't
4203 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
4205 * config/i386/i386.h (enum reg_class): Rename MASK_REGS to
4206 ALL_MASK_REGS and MASK_EVEX_REGS to MASK_REGS.
4207 (MASK_CLASS_P): Update for rename.
4208 (MAYBE_MASK_CLASS_P): Ditto.
4209 (REG_CLASS_NAMES): Update.
4210 (REG_CLASS_CONTENT): Update.
4211 * config/i386/i386.c (regclass_map): Update for MASK_REG
4212 and ALL_MASK_REGS rename.
4213 * config/i386/constraints.md (Yk): Update for rename.
4216 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
4218 * config/i386/i386.h (enum reg_class): Remove
4219 EVEX_SSE_REGS and MOD4_SSE_REGS.
4220 (REG_CLASS_NAMES): Update.
4221 (REG_CLASS_CONTENT): Update.
4222 * config/i386/i386.c (regclass_map): Declare AVX-512 SSE
4223 registers as ALL_SSE_REGS.
4224 (ix86_additional_allocno_class_p): Remove.
4225 (TARGET_ADDITIONAL_ALLOCNO_CLASS_P): Remove.
4226 (ix86_register_priority): Lower priority of EVEX SSE registers.
4227 Use IN_RANGE macro where appropriate.
4228 (ix86_hard_regno_mode_ok): Merge AVX-5124FMAPS and
4229 AVX-5124VNNIW checks.
4230 (ix86_modes_tieable_p): Tie 512-bit SSE modes.
4231 * config/i386/sse.md (avx5124fmaddps_4fmaddps)
4232 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddps_maskz)
4233 (avx5124fmaddps_4fmaddss, avx5124fmaddps_4fmaddss_mask)
4234 (avx5124fmaddps_4fmaddss_maskz, avx5124fmaddps_4fnmaddps)
4235 (avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddps_maskz)
4236 (avx5124fmaddps_4fnmaddss, avx5124fmaddps_4fnmaddss_mask)
4237 (avx5124fmaddps_4fnmaddss_maskz, avx5124vnniw_vp4dpwssd)
4238 (avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssd_maskz)
4239 (avx5124vnniw_vp4dpwssds, avx5124vnniw_vp4dpwssds_mask)
4240 (avx5124vnniw_vp4dpwssds_maskz): Use "v" instead of "Yh" constraint.
4241 * config/i386/constraints.md (Yh): Remove.
4243 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
4245 * config/i386/i386.c (regclass_map): Declare integer REX registers
4248 2018-09-23 Gerald Pfeifer <gerald@pfeifer.com>
4250 * doc/service.texi (Service): Switch the fsf.org link to https.
4252 2018-09-22 Chung-Ju Wu <jasonwucj@gmail.com>
4255 * config/nds32/nds32.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
4256 Define to speculation_safe_value_not_needed.
4258 2018-09-21 Florian Weimer <fweimer@redhat.com>
4261 * doc/extend.texi (Common Function Attributes): Mention that
4262 noreturn suppresses tail call optimization.
4264 2018-09-21 Jeff Law <law@redhat.com>
4266 * gimple-ssa-evrp.c (evrp_dom_walker::cleanup): Call
4267 vr_values::cleanup_edges_and_switches.
4268 * tree-vrp.c (to_remove_edges, to_update_switch_stmts): Moved into
4270 (identify_jump_threads): Remove EDGE_IGNORE handling.
4271 (execute_vrp): Move handling of to_remove_edges and
4272 to_update_switch_stmts into vr_values class member functions.
4273 * tree-vrp.h (switch_update, to_remove_edges): Remove declarations.
4274 (to_update_switch_stmts): Likewise.
4275 * vr-values.c: Include cfghooks.h.
4276 (vr_values::vr_values): Initialize to_remove_edges and
4277 to_update_switch_stmts.
4278 (vr_values::~vr_values): Verify to_remove_edges and
4279 to_update_switch_stmts are empty.
4280 (vr_values::simplify_switch_using_ranges): Set EDGE_IGNORE as needed.
4281 (vr_values::cleanup_edges_and_switches): New member function.
4282 * vr-values.h (vr_values): Add cleanup_edges_and_switches member
4283 function. Add new data members.
4285 2018-09-21 David Malcolm <dmalcolm@redhat.com>
4287 PR tree-optimization/87309
4288 * dumpfile.c (dump_context::begin_scope): Filter the dump_loc
4289 calls with pflags and alt_flags.
4290 (selftest::test_capture_of_dump_calls): Add test of interaction of
4291 MSG_OPTIMIZED_LOCATIONS with AUTO_DUMP_SCOPE.
4293 2018-09-21 Olivier Hainque <hainque@adacore.com>
4295 * config.gcc: Factorize and comment inclusion of vxworks-dummy.h.
4297 2018-09-21 Olivier Hainque <hainque@adacore.com>
4299 * config/vxworks.h (CLEAR_INSN_CACHE): #define to 1.
4301 2018-09-21 Olivier Hainque <hainque@adacore.com>
4303 * config/vxworks.h (VXWORKS_LIBGCC_SPEC): Remove -lc_internal.
4304 Merge block comment with the one ahead of VXWORKS_LIBS_RTP. Then:
4305 (VXWORKS_LIBS_RTP): Minor reordering.
4307 2018-09-21 Olivier Hainque <hainque@adacore.com>
4309 * config/vxworks.h (STARTFILE_PREFIX_SPEC): Define.
4310 (VXWORKS_LIBS_DIR_RTP): Remove definition and use.
4312 2018-09-21 Olivier Hainque <hainque@adacore.com>
4314 * config/vxworks.h (SIZE_TYPE): Account for TARGET_VXWORKS64.
4315 (PTRDIFF_TYPE): Likewise.
4317 2018-09-21 Olivier Hainque <hainque@adacore.com>
4319 * config.gcc: Enforce def of TARGET_VXWORKS64 to 1 from
4320 triplet, similar to support for VxWorks7.
4321 * config/vxworks-dummy.h: Provide a default definition
4322 of TARGET_VXWORKS64 to 0.
4324 2018-09-21 Olivier Hainque <hainque@adacore.com>
4326 * config/vxworks.h (TARGET_VXWORKS7): Move default definition ...
4327 * config/vxworks-dummy.h: here.
4329 2018-09-21 Olivier Hainque <hainque@adacore.com>
4331 * config.gcc: Prepend vxworks-dummy.h to tm_file for powerpc*
4333 2018-09-21 Shaokun Zhang <zhangshaokun@hisilicon.com>
4334 Bo Zhou <zbo.zhou@hisilicon.com>
4336 * config/aarch64/aarch64-cores.def (tsv110): New CPU.
4337 * config/aarch64/aarch64-tune.md: Regenerated.
4338 * doc/invoke.texi (AArch64 Options/-mtune): Add "tsv110".
4339 * config/aarch64/aarch64.c (tsv110_tunings): New tuning table.
4340 * config/aarch64/aarch64-cost-tables.h: Add "tsv110" extra costs.
4342 2018-09-21 Andrew Stubbs <ams@codesourcery.com>
4343 Julian Brown <julian@codesourcery.com>
4345 * builtins.c (get_builtin_sync_mem): Handle address spaces.
4347 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
4349 * config/rs6000/rs6000.c (rs6000_function_ok_for_sibcall): Return false
4350 if the call takes a static chain.
4352 2018-09-21 Martin Liska <mliska@suse.cz>
4354 * auto-profile.c (autofdo_source_profile::read): Do not
4356 (read_profile): Do not add working sets.
4357 (read_autofdo_file): Remove sum_all.
4358 (afdo_callsite_hot_enough_for_early_inline): Remove const
4360 * coverage.c (struct counts_entry): Remove gcov_summary.
4361 (read_counts_file): Read new GCOV_TAG_OBJECT_SUMMARY,
4362 do not support GCOV_TAG_PROGRAM_SUMMARY.
4363 (get_coverage_counts): Remove summary and expected
4365 * coverage.h (get_coverage_counts): Likewise.
4366 * doc/gcov-dump.texi: Remove -w option.
4367 * gcov-dump.c (dump_working_sets): Remove.
4368 (main): Do not support '-w' option.
4369 (print_usage): Likewise.
4370 (tag_summary): Likewise.
4371 * gcov-io.c (gcov_write_summary): Do not dump
4373 (gcov_read_summary): Likewise.
4374 (gcov_histo_index): Remove.
4375 (gcov_histogram_merge): Likewise.
4376 (compute_working_sets): Likewise.
4377 * gcov-io.h (GCOV_TAG_OBJECT_SUMMARY): Mark
4379 (GCOV_TAG_PROGRAM_SUMMARY): Mark it obsolete.
4380 (GCOV_TAG_SUMMARY_LENGTH): Adjust.
4381 (GCOV_HISTOGRAM_SIZE): Remove.
4382 (GCOV_HISTOGRAM_BITVECTOR_SIZE): Likewise.
4383 (struct gcov_summary): Simplify rapidly just
4384 to runs and sum_max fields.
4385 (gcov_histo_index): Remove.
4386 (NUM_GCOV_WORKING_SETS): Likewise.
4387 (compute_working_sets): Likewise.
4388 * gcov-tool.c (print_overlap_usage_message): Remove
4389 trailing empty line.
4390 * gcov.c (read_count_file): Read GCOV_TAG_OBJECT_SUMMARY.
4391 (output_lines): Remove program related line.
4392 * ipa-profile.c (ipa_profile): Do not consider GCOV histogram.
4393 * lto-cgraph.c (output_profile_summary): Do not stream GCOV
4395 (input_profile_summary): Do not read it.
4396 (merge_profile_summaries): And do not merge it.
4397 (input_symtab): Do not call removed function.
4398 * modulo-sched.c (sms_schedule): Do not print sum_max.
4399 * params.def (HOT_BB_COUNT_FRACTION): Reincarnate param that was
4400 removed when histogram method was invented.
4401 (HOT_BB_COUNT_WS_PERMILLE): Mention that it's used only in LTO
4403 * postreload-gcse.c (eliminate_partially_redundant_load): Fix
4405 * predict.c (get_hot_bb_threshold): Use HOT_BB_COUNT_FRACTION
4406 and dump selected value.
4407 * profile.c (add_working_set): Remove.
4408 (get_working_sets): Likewise.
4409 (find_working_set): Likewise.
4410 (get_exec_counts): Do not work with working sets.
4411 (read_profile_edge_counts): Do not inform as sum_max is removed.
4412 (compute_branch_probabilities): Likewise.
4413 (compute_value_histograms): Remove argument for call of
4414 get_coverage_counts.
4415 * profile.h: Do not make gcov_summary const.
4417 2018-09-21 Monk Chiang <sh.chiang04@gmail.com>
4419 * config.gcc (nds32*-*-*): Set TARGET_DEFAULT_TLSDESC_TRAMPOLINE=0.
4421 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
4423 PR tree-optimization/86990
4424 * gimple-ssa-store-merging.c (imm_store_chain_info:coalesce_immediate):
4425 Check that the entire merged store group is made of constants only for
4428 2018-09-20 Allan Sandfeld Jensen <allan.jensen@qt.io>
4430 * gcc.c (LINK_COMMAND_SPEC): Handle -r like -nostdlib.
4431 (VTABLE_VERIFICATION_SPEC): Likewise.
4432 (SANITIZER_EARLY_SPEC): Likewise.
4433 (SANITIZER_SPEC): Likewise.
4434 * config/darwin.h (LINK_COMMAND_SPEC): Likewise.
4435 * doc/invoke.texi (Link Options): Document -r.
4437 2018-09-20 Richard Biener <rguenther@suse.de>
4440 * gimplify.c (gimplify_expr): Retain alignment of
4441 addressable lvalue in dereference.
4443 2018-09-20 Alexandre Oliva <aoliva@redhat.com>
4446 * configure.ac: Check for .loc is_stmt support.
4447 * configure, config.in: Rebuilt.
4448 * dwarf2out.c (dwarf2out_source_line): Skip is_stmt
4451 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
4453 * config/rs6000/rs6000.opt (misel=no, misel=yes): Delete.
4454 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -misel=yes and
4457 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
4459 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
4461 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete
4464 2018-09-20 Marek Polacek <polacek@redhat.com>
4466 * doc/invoke.texi: Add -Wno-init-list-lifetime to C++ Language Options.
4468 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
4470 PR tree-optimization/87288
4471 * tree-vect-loop.c (vect_analyze_loop_2): Take PEELING_FOR_GAPS
4472 into account when determining PEELING_FOR_NITERS.
4474 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
4476 PR tree-optimization/86877
4477 * tree-vect-loop.c (vect_analyze_loop_2): Call
4478 vect_verify_datarefs_alignment.
4480 2018-09-19 Marek Polacek <polacek@redhat.com>
4482 * doc/invoke.texi: Document -Wclass-conversion.
4484 2018-09-19 John David Anglin <danglin@gcc.gnu.org>
4486 * config/pa/pa.c (pa_adjust_priority): Delete.
4487 (TARGET_SCHED_ADJUST_PRIORITY): Delete define.
4489 * config/pa/pa.md (atomic_storeqi): Restore deleted expander.
4490 (atomic_storehi): Likewise.
4491 (atomic_storesi): Likewise.
4492 (atomic_loaddi): Restore compare and swap exchange loop code.
4494 2018-09-19 Segher Boessenkool <segher@kernel.crashing.org>
4496 PR rtl-optimization/86902
4497 * combine.c (try_combine): When changing the CC mode used, don't change
4498 an unrelated mode in other_insn to that new CC mode.
4500 2018-09-19 David Malcolm <dmalcolm@redhat.com>
4502 * tree-data-ref.c (runtime_alias_check_p): Use formatted printing
4503 with %T in place of calls to dump_generic_expr.
4504 (prune_runtime_alias_test_list): Likewise.
4505 (create_runtime_alias_checks): Likewise.
4506 * tree-vect-data-refs.c (vect_check_nonzero_value): Likewise.
4507 (vect_analyze_data_ref_dependence): Likewise.
4508 (vect_slp_analyze_data_ref_dependence): Likewise.
4509 (vect_record_base_alignment): Likewise. Use %G in place of call
4510 to dump_gimple_stmt.
4511 (vect_compute_data_ref_alignment): Likewise.
4512 (verify_data_ref_alignment): Likewise.
4513 (vect_find_same_alignment_drs): Likewise.
4514 (vect_analyze_group_access_1): Likewise.
4515 (vect_analyze_data_ref_accesses): Likewise.
4516 (dependence_distance_ge_vf): Likewise.
4517 (dump_lower_bound): Likewise.
4518 (vect_prune_runtime_alias_test_list): Likewise.
4519 (vect_find_stmt_data_reference): Likewise.
4520 (vect_analyze_data_refs): Likewise.
4521 (vect_create_addr_base_for_vector_ref): Likewise.
4522 (vect_create_data_ref_ptr): Likewise.
4523 * tree-vect-loop-manip.c (vect_set_loop_condition): Likewise.
4524 (vect_can_advance_ivs_p): Likewise.
4525 (vect_update_ivs_after_vectorizer): Likewise.
4526 (vect_gen_prolog_loop_niters): Likewise.
4527 (vect_prepare_for_masked_peels): Likewise.
4528 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
4529 (vect_determine_vectorization_factor): Likewise.
4530 (vect_is_simple_iv_evolution): Likewise.
4531 (vect_analyze_scalar_cycles_1): Likewise.
4532 (vect_analyze_loop_operations): Likewise.
4533 (report_vect_op): Likewise.
4534 (vect_is_slp_reduction): Likewise.
4535 (check_reduction_path): Likewise.
4536 (vect_is_simple_reduction): Likewise.
4537 (vect_create_epilog_for_reduction): Likewise.
4538 (vect_finalize_reduction:): Likewise.
4539 (vectorizable_induction): Likewise.
4540 (vect_transform_loop_stmt): Likewise.
4541 (vect_transform_loop): Likewise.
4542 (optimize_mask_stores): Likewise.
4543 * tree-vect-patterns.c (vect_pattern_detected): Likewise.
4544 (vect_split_statement): Likewise.
4545 (vect_recog_over_widening_pattern): Likewise.
4546 (vect_recog_average_pattern): Likewise.
4547 (vect_determine_min_output_precision_1): Likewise.
4548 (vect_determine_precisions_from_range): Likewise.
4549 (vect_determine_precisions_from_users): Likewise.
4550 (vect_mark_pattern_stmts): Likewise.
4551 (vect_pattern_recog_1): Likewise.
4552 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
4553 (vect_record_max_nunits): Likewise.
4554 (vect_build_slp_tree_1): Likewise.
4555 (vect_build_slp_tree_2): Likewise.
4556 (vect_print_slp_tree): Likewise.
4557 (vect_analyze_slp_instance): Likewise.
4558 (vect_detect_hybrid_slp_stmts): Likewise.
4559 (vect_detect_hybrid_slp_1): Likewise.
4560 (vect_slp_analyze_operations): Likewise.
4561 (vect_slp_analyze_bb_1): Likewise.
4562 (vect_transform_slp_perm_load): Likewise.
4563 (vect_schedule_slp_instance): Likewise.
4564 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
4565 (vect_mark_stmts_to_be_vectorized): Likewise.
4566 (vect_init_vector_1): Likewise.
4567 (vect_get_vec_def_for_operand): Likewise.
4568 (vect_finish_stmt_generation_1): Likewise.
4569 (vect_check_load_store_mask): Likewise.
4570 (vectorizable_call): Likewise.
4571 (vectorizable_conversion): Likewise.
4572 (vectorizable_operation): Likewise.
4573 (vectorizable_load): Likewise.
4574 (vect_analyze_stmt): Likewise.
4575 (vect_is_simple_use): Likewise.
4576 (vect_get_vector_types_for_stmt): Likewise.
4577 (vect_get_mask_type_for_stmt): Likewise.
4578 * tree-vectorizer.c (increase_alignment): Likewise.
4580 2018-09-19 Andrew Stubbs <ams@codesourcery.com>
4582 * doc/rtl.texi: Adjust vec_select description.
4583 * simplify-rtx.c (simplify_binary_operation_1): Allow VEC_SELECT to use
4584 non-constant selectors.
4586 2018-09-19 Matthew Malcomson <matthew.malcomson@arm.com>
4588 * config/aarch64/aarch64-protos.h
4589 (aarch64_offset_9bit_signed_unscaled_p): New declaration.
4590 * config/aarch64/aarch64.md (arches): New "rcpc8_4" attribute value.
4591 (arch_enabled): Add check for "rcpc8_4" attribute value of "arch".
4592 * config/aarch64/aarch64.h (AARCH64_FL_RCPC8_4): New bitfield.
4593 (AARCH64_FL_FOR_ARCH8_4): Include AARCH64_FL_RCPC8_4.
4594 (AARCH64_FL_PROFILE): Move index so flags are ordered.
4595 (AARCH64_ISA_RCPC8_4): New flag.
4596 * config/aarch64/aarch64.c (offset_9bit_signed_unscaled_p): Renamed
4597 to aarch64_offset_9bit_signed_unscaled_p.
4598 * config/aarch64/atomics.md (atomic_store<mode>): Allow offset
4600 * config/aarch64/constraints.md (Ust): New constraint.
4601 * config/aarch64/predicates.md.
4602 (aarch64_9bit_offset_memory_operand): New predicate.
4603 (aarch64_rcpc_memory_operand): New predicate.
4605 2018-09-19 Eric Botcazou <ebotcazou@adacore.com>
4607 PR rtl-optimization/87361
4608 * rtlanal.c (nonzero_bits1): Revert accidental change.
4610 2018-09-19 Richard Biener <rguenther@suse.de>
4612 PR tree-optimization/87349
4613 PR tree-optimization/87342
4614 * tree-ssa-sccvn.c (do_rpo_vn): Iterate max_rpo computation.
4616 2018-09-18 Marek Polacek <polacek@redhat.com>
4618 P1064R0 - Allowing Virtual Function Calls in Constant Expressions
4619 * gimple-fold.c (gimple_get_virt_method_for_vtable): Adjust assert.
4621 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
4623 * config/rs6000/rs6000.md: Remove old "Cygnus sibcall" comment.
4625 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
4627 PR rtl-optimization/86882
4628 * rtlanal.c (reg_overlap_mentioned_p): Handle CLOBBER.
4630 2018-09-18 Uros Bizjak <ubizjak@gmail.com>
4632 * config/i386/i386.md (*<code>extend<mode>xf2): Macroize insn from
4633 *<code>extendsfxf2 and *<code>extenddfxf2 using MODEF mode iterator.
4635 2018-09-18 Jonathan Wakely <jwakely@redhat.com>
4638 * doc/invoke.texi (Link Options): Fix formatting and grammar.
4640 2018-09-18 Richard Biener <rguenther@suse.de>
4643 * tree-ssa-coalesce.c (tree_int_map_hasher): Remove.
4644 (compute_samebase_partition_bases): Likewise.
4645 (coalesce_ssa_name): Always use compute_optimized_partition_bases.
4646 (gimple_can_coalesce_p): Simplify.
4648 2018-09-18 Hans-Peter Nilsson <hp@bitrange.com>
4650 Handle a library implementation of ffs calling __builtin_ffs.
4651 * config/mmix/mmix.c (TARGET_INIT_LIBFUNCS): Override with...
4652 (mmix_init_libfuncs): New function: make __builtin_ffs expand
4655 2018-09-17 David Malcolm <dmalcolm@redhat.com>
4657 * diagnostic-show-locus.c (class layout_range): Add field
4659 (layout_range::layout_range): Add "original_idx" param and use it
4660 to initialize new field.
4661 (make_range): Use 0 for original_idx.
4662 (layout::layout): Pass in index to calls to
4663 maybe_add_location_range.
4664 (layout::maybe_add_location_range): Add param "original_idx" and
4665 pass it on to layout_range.
4666 (layout::print_any_labels): Pass on range->m_original_idx to
4668 (gcc_rich_location::add_location_if_nearby): Use 0 for
4670 * gcc-rich-location.h (text_range_label::get_text): Update for new
4672 (range_label_for_type_mismatch::get_text): Likewise.
4674 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
4676 * config/i386/i386.c (ix86_emit_i387_log1p): Emit fldln2 earlier.
4678 2018-09-17 David Malcolm <dmalcolm@redhat.com>
4680 * gimple-ssa-sprintf.c (fmtwarn): Update for introduction of
4681 format_string_diagnostic_t.
4682 (fmtwarn_n): Likewise.
4683 * substring-locations.c
4684 (format_string_diagnostic_t::format_string_diagnostic_t) New ctor.
4685 (format_warning_n_va): Convert to...
4686 (format_string_diagnostic_t::emit_warning_n_va): ...this.
4687 (format_warning_va): Convert to...
4688 (format_string_diagnostic_t::emit_warning_va): ...this.
4689 (format_warning_at_substring): Convert to...
4690 (format_string_diagnostic_t::emit_warning): ...this.
4691 (format_warning_at_substring_n): Convert to...
4692 (format_string_diagnostic_t::emit_warning_n): ...this.
4693 * substring-locations.h (class format_string_diagnostic_t): New
4695 (format_warning_va): Convert to
4696 format_string_diagnostic_t::emit_warning_va.
4697 (format_warning_n_va): Convert to
4698 format_string_diagnostic_t::emit_warning_n_va.
4699 (format_warning_at_substring): Convert to
4700 format_string_diagnostic_t::emit_warning.
4701 (format_warning_at_substring_n): Convert to
4702 format_string_diagnostic_t::emit_warning_n.
4704 2018-09-17 Cesar Philippidis <cesar@codesourcery.com>
4705 Bernd Schmidt <bernds_cb1@t-online.de>
4707 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with
4710 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
4712 * config/i386/i386.md (truncxf<mode>2_i387_noop_unspec): Change
4713 operand 0 predicate to nonimmediate operand.
4714 (rint<mode>2_frndint): Remove insn pattern.
4715 (rint<mode>2): Change operand 1 predicate to general_operand.
4716 Extend operand 1 to XFmode and generate rintxf2 insn.
4717 (frndintxf2_<rounding>): Rename from frndint<mode>2_<rounding>.
4718 Do not use X87MODEF mode macro.
4719 (frndintxf2_<rounding>_i387): Rename from
4720 frndint<mode>2_<rounding>_i387. Do not use X87MODEF mode macro.
4721 (<rounding_insn><mode>2): For non-SSE modes, extend operand 1
4722 to XFmode and generate significandxf3 insn.
4724 2018-09-17 Richard Biener <rguenther@suse.de>
4726 PR tree-optimization/87328
4727 * tree-ssa-sccvn.c (process_bb): Remove assertion about not
4728 visiting unexecutable backedges when not iterating.
4729 (do_rpo_vn): Mark all edges not executable even when not
4732 2018-09-17 Martin Jambor <mjambor@suse.cz>
4735 * doc/invoke.texi (Warning Options): Likewise.
4737 2018-09-17 Richard Biener <rguenther@suse.de>
4739 PR tree-optimization/87301
4740 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup): Properly
4741 clean EH info from leftover copy assignments.
4743 2018-09-17 Martin Liska <mliska@suse.cz>
4745 PR gcov-profile/85871
4746 * gcov.c (output_intermediate_file): Fix out of bounds
4749 2018-09-17 Vineet Gupta <vgupta@synopsys.com>
4751 * config/arc/arc.c: Object attributes for core4 not reflected
4753 * config/arc/arc.h: Don't restrict DBNZ to core3 (core4 includes
4756 2018-09-17 Alexey Brodkin <abrodkin@synopsys.com>
4758 * config/arc/linux.h (LINK_EH_SPEC): Add missing space.
4760 2018-09-17 Martin Liska <mliska@suse.cz>
4762 * doc/gcov.texi: Document new option --use-hotness-colors.
4763 * gcov.c (struct source_info): Declare new field.
4764 (source_info::source_info): Set default for maximum_count.
4765 (print_usage): Add new -q option.
4766 (process_args): Process it.
4767 (accumulate_line_info): Save src->maximum_count.
4768 (output_line_beginning): Make color line number if
4769 flag_use_hotness_colors is set.
4770 (output_line_details): Pass default argument value.
4771 (output_lines): Pass src->maximum_count.
4773 2018-09-17 Martin Liska <mliska@suse.cz>
4775 * common/config/i386/i386-common.c (ix86_get_valid_option_values):
4776 Use processor_names table.
4777 * config/i386/i386.c (ix86_default_align): Use
4778 processor_cost_table for alignment values.
4779 (ix86_option_override_internal): Use processor_names.
4780 (ix86_function_specific_print): Likewise.
4781 * config/i386/i386.h (struct processor_costs):
4782 Add alignment values.
4783 (struct ptt): Remove and replace with const char *.
4784 * config/i386/x86-tune-costs.h (struct processor_costs):
4785 Declare default alignments for all costs.
4787 2018-09-17 Aldy Hernandez <aldyh@redhat.com>
4789 * tree-vrp.c (extract_range_from_unary_expr): Do not special case
4790 symbolics or VR_VARYING ranges for ABS_EXPR.
4791 * wide-int-range.cc (wide_int_range_abs): Return positive numbers
4792 when range will wrap.
4794 2018-09-15 Eric Botcazou <ebotcazou@adacore.com>
4797 * cfgexpand.c (expand_gimple_basic_block): Be prepared for a BARRIER
4798 before and after a JUMP_TABLE_DATA.
4800 2018-09-14 John David Anglin <danglin@gcc.gnu.org>
4803 * dojump.c (do_compare_and_jump): Canonicalize function pointers
4804 when one operand is a function pointer. Use POINTER_TYPE_P and
4805 FUNC_OR_METHOD_TYPE_P.
4806 * expr.c (do_store_flag): Use POINTER_TYPE_P and FUNC_OR_METHOD_TYPE_P.
4807 * fold-const.c (build_range_check): Likewise.
4808 * match.pd (simple_comparison): Likewise.
4810 2018-09-14 David Malcolm <dmalcolm@redhat.com>
4813 * spellcheck.c (get_edit_distance_cutoff): New function.
4814 (selftest::test_edit_distance_unit_test_oneway): Rename to...
4815 (selftest::test_get_edit_distance_one_way): ...this.
4816 (selftest::test_get_edit_distance_unit): Rename to...
4817 (selftest::test_get_edit_distance_both_ways): ...this.
4818 (selftest::test_edit_distances): Move tests to this new function,
4819 and test some more pairs of strings. Update for above renaming.
4820 (selftest::get_old_cutoff): New function.
4821 (selftest::test_get_edit_distance_cutoff): New function.
4822 (selftest::assert_suggested_for): New function.
4823 (ASSERT_SUGGESTED_FOR): New macro.
4824 (selftest::assert_not_suggested_for): New function.
4825 (ASSERT_NOT_SUGGESTED_FOR): New macro.
4826 (selftest::test_suggestions): New function.
4827 (selftest::spellcheck_c_tests): Move test_get_edit_distance_unit
4828 tests to selftest::test_edit_distances and call it. Add calls to
4829 selftest::test_get_edit_distance_cutoff and
4830 selftest::test_suggestions.
4831 * spellcheck.h (get_edit_distance_cutoff): New function declaration.
4832 (best_match::consider): Replace hard-coded cutoff calculation with
4834 (best_match::get_cutoff): New declaration.
4835 (best_match::get_best_meaningful_candidate): Likewise.
4837 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
4839 * builtins.c (fold_builtin_strlen): Remove TODO comment.
4841 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
4844 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
4846 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
4847 terminated string literal.
4849 2018-09-14 Martin Sebor <msebor@redhat.com>
4851 * builtins.c (unterminated_array): Handle ARRAY_REF.
4852 (expand_builtin_stpcpy_1): Detect unterminated char arrays.
4853 * builtins.h (unterminated_array): Declare extern.
4854 * gimple-fold.c (gimple_fold_builtin_stpcpy): Detect unterminated
4856 (gimple_fold_builtin_sprintf): Propagate NO_WARNING to transformed
4859 2018-09-14 Martin Sebor <msebor@redhat.com>
4860 Jeff Law <law@redhat.com>
4862 * builtins.c (unterminated_array): New.
4863 (expand_builtin_strcpy): Adjust.
4864 (expand_builtin_strcpy_args): Detect unterminated arrays.
4865 * gimple-fold.c (get_maxval_strlen): Add argument. Detect
4866 unterminated arrays.
4867 * gimple-fold.h (get_maxval_strlen): Add argument.
4868 (gimple_fold_builtin_strcpy): Detec unterminated arrays.
4870 * gimple-fold.c (get_range_strlen): Add argument.
4871 (get_maxval_strlen): Adjust.
4872 * gimple-fold.h (get_range_strlen): Add argument.
4874 2018-09-14 Wei Xiao <wei3.xiao@intel.com>
4876 * config/i386/movdirintrin.h: Fix copyright year.
4878 2018-09-14 Uros Bizjak <ubizjak@gmail.com>
4880 * reg-stack.c: Include regs.h.
4881 (replace_reg): Assert that mode is MODE_FLOAT or MODE_COMPLEX_FLOAT.
4882 (emit_pop_insn): Default pop insn mode to the reg_raw_mode of
4883 FIRST_STACK_REG, not DFmode.
4884 (emit_swap_insn): Default swap insn mode to the reg_raw_mode of
4885 FIRST_STACK_REG, not XFmode. Explicitly construct swap RTX.
4886 (change stack): Default register mode to the reg_raw_mode of
4887 FIRST_STACK_REG, not DFmode.
4888 * config/i386/i386.md (*swap<mode>): Remove insn pattern.
4889 (*swapxf): Rename from swapxf.
4891 2018-09-14 Carl Love <cel@us.ibm.com>
4893 * config/rs6000/emmintrin.h: Add _MM_SHUFFLE2.
4894 * config/rs6000/xmmintrin.h: Add _MM_SHUFFLE.
4896 2018-09-14 Segher Boessenkool <segher@kernel.crashing.org>
4899 * config/rs6000/rs6000.md (*mov<mode>_hardfloat64): Add Z to the Y
4902 2018-09-14 Sam Tebbs <sam.tebbs@arm.com>
4905 * config/aarch64/aarch64.md (*aarch64_bfxilsi_uxtw): Define.
4907 2018-09-14 Jason Merrill <jason@redhat.com>
4909 Fix --enable-gather-detailed-mem-stats.
4910 * hash-table.c (hash_table_usage): Change from variable to function.
4911 * hash-table.h: Adjust.
4912 * Makefile.in: Add missing dependencies on hash-table.h.
4914 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4916 PR tree-optimization/87259
4918 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
4919 execute_cse_reciprocals_1 has tried transforming.
4921 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
4923 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
4924 VR_VARYING for PLUS/MINUS_EXPR.
4926 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
4928 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
4931 2018-09-14 Richard Biener <rguenther@suse.de>
4934 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
4935 bits for the merged partition.
4937 2018-09-13 Martin Sebor <msebor@redhat.com>
4938 Bernd Edlinger <bernd.edlinger@hotmail.de>
4940 * builtins.h (c_srlen): Add argument.
4941 * builtins.c (warn_string_no_nul): New function.
4942 (c_strlen): Add argument and use it. Update recursive calls.
4943 Pass DECL argument to string_constant to get info on non
4944 terminated strings. Update *NONSTR as needed.
4945 (fold_builtin_strlen): Add argument to calls to c_strlen.
4946 Warn for unterminated arrays.
4947 (warn_string_no_null): Add prototype.
4948 * expr.c (string_constant): Update arguments. Update recursive
4949 calls appropriately. Detect missing NUL terminator and outermost
4950 declaration its missing in.
4951 Improve checks for arrays with nonzero lower bound or elements
4952 that are not a single byte. Simplify offset computation.
4953 Simplify checks for non-NUL terminated strings.
4954 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
4955 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
4957 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
4959 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
4961 * fold-const.c (c_getstr): Fix function comment. Remove unused third
4962 argument. Fix range checks.
4963 * fold-const.h (c_getstr): Adjust protoype.
4964 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
4965 string is constant but contains no NUL byte.
4967 * expr.c (string_constant): Adjust function comment.
4968 Remove bogus check for zero termination.
4970 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
4972 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
4973 (get_constant_size): Don't make STRING_CSTs larger than they are.
4974 (check_string_literal): New check function for STRING_CSTs.
4975 (output_constant): Use it.
4977 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
4980 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
4982 2018-09-13 Richard Biener <rguenther@suse.de>
4984 PR tree-optimization/87263
4985 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
4986 (struct unwind_state): Add max_rpo field.
4987 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
4988 Compute max_rpo, the max RPO number a block can be backwards reached
4989 from. Re-write non-iterating mode to a RPO ordered worklist approach,
4990 separating it from the iterating mode.
4992 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
4994 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
4995 (rfs_decision): New scheduling decision.
4997 2018-09-13 Richard Biener <rguenther@suse.de>
5000 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
5001 (vn_nary_op_insert_pieces_predicated): Do not write useless
5002 valid_dominated_by_p entry outside of the allocated storage.
5004 2018-09-13 Omar Sandoval <osandov@osandov.com>
5005 Tom de Vries <tdevries@suse.de>
5008 * dwarf2out.c (is_c): New function.
5009 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
5011 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
5014 * config/aarch64/aarch64.md (*aarch64_bfxil):
5016 * config/aarch64/constraints.md (Ulc): Define.
5017 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
5019 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
5022 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
5024 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
5025 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
5026 aarch64_layout_frame call.
5027 (aarch64_expand_epilogue): Likewise.
5028 (aarch64_initial_elimination_offset): Likewise.
5029 (aarch64_get_separate_components): Likewise.
5030 (aarch64_use_return_insn_p): Likewise.
5031 (aarch64_layout_frame): Remove unneeded check.
5033 2018-09-13 Jakub Jelinek <jakub@redhat.com>
5035 * configure.ac: Only append
5036 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
5037 gcc_config_arguments if it was never reconfigured or last reconfigure
5038 was with different arguments.
5039 * configure: Regenerated.
5041 2018-09-13 Jakub Jelinek <jakub@redhat.com>
5042 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5045 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
5046 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
5048 2018-09-13 Jakub Jelinek <jakub@redhat.com>
5050 PR tree-optimization/87287
5051 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
5052 X % C == 0 to X % (unsigned) C == 0 optimization to ...
5053 * match.pd (X % C == 0): ... here. New optimization.
5055 2018-09-12 Jakub Jelinek <jakub@redhat.com>
5058 * expr.h (maybe_optimize_mod_cmp): Declare.
5059 * expr.c (mod_inv): New function.
5060 (maybe_optimize_mod_cmp): New function.
5061 (do_store_flag): Use it.
5062 * cfgexpand.c (expand_gimple_cond): Likewise.
5064 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
5065 Julian Brown <julian@codesourcery.com>
5068 * gimplify.c (gimplify_scan_omp_clauses): Set
5069 target_firstprivatize_array_bases in OpenACC parallel and kernels
5070 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
5071 OpenACC data regions.
5073 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
5075 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
5076 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
5078 2018-09-12 Richard Biener <rguenther@suse.de>
5080 PR tree-optimization/87280
5081 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
5082 edge but unreachable target.
5083 (do_rpo_vn): For conservatively handling a PHI only mark
5084 the backedge executable but not the block reachable.
5086 2018-09-12 Richard Biener <rguenther@suse.de>
5088 PR tree-optimization/87266
5089 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
5092 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
5094 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
5096 ("trunc<BFP:mode><DFP_ALL:mode>2")
5097 ("trunc<DFP_ALL:mode><BFP:mode>2")
5098 ("extend<BFP:mode><DFP_ALL:mode>2")
5099 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
5100 according to the target operand type.
5102 2018-09-12 Jakub Jelinek <jakub@redhat.com>
5103 Andreas Krebbel <krebbel@linux.ibm.com>
5105 PR tree-optimization/86844
5106 * gimple-ssa-store-merging.c
5107 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
5108 there are any overlapping stores in between them, make sure they are
5109 also coalesced or we give up completely.
5111 2018-09-12 Jakub Jelinek <jakub@redhat.com>
5114 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
5115 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
5117 2018-09-12 Tom de Vries <tdevries@suse.de>
5119 * common.opt (gdescribe-dies): Add option.
5120 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
5121 attribute for artifical and nameless decls.
5122 (dwarf2out_register_external_die): Add description attribute to
5123 external reference die.
5124 (add_desc_attribute): New functions.
5125 (gen_subprogram_die): Add description attribute to
5126 DW_TAG_call_site_parameter.
5127 * tree-pretty-print.c (print_generic_expr_to_str): New function.
5128 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
5129 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
5131 (@item -gdescribe-dies): Add.
5133 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
5135 * tree-vrp.c (vrp_shift_undefined_p): Remove.
5136 (extract_range_from_binary_expr_1: Call
5137 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
5138 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
5141 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
5143 * gimple-ssa-warn-alloca.c
5144 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
5145 field for ALLOCA_BOUND_*_LARGE.
5147 2018-09-11 Nathan Sidwell <nathan@acm.org>
5149 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
5151 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
5153 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
5154 for clobbers. Remove obsolete comment.
5156 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
5158 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
5159 mpxchk, mpxld and mpxst types.
5160 (define_attr length_immediate): Remove all processing of mpx types.
5161 (define_attr prefix_0f): Ditto.
5162 (define_attr memory): Ditto.
5164 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
5166 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
5167 (log<mode>2): Change operand 1 predicate to general_operand.
5168 Extend operand 1 to XFmode and generate logxf3 insn.
5169 (log10<mode>2): Change operand 1 predicate to general_operand.
5170 Extend operand 1 to XFmode and generate log10xf3 insn.
5171 (log2<mode>2): Change operand 1 predicate to general_operand.
5172 Extend operand 1 to XFmode and generate log2xf3 insn.
5173 (fyl2xp1_extend<mode>xf3_i387): Remove.
5174 (log1p<mode>2): Change operand 1 predicate to general_operand.
5175 Extend operand 1 to XFmode and generate log1pxf3 insn.
5176 (fxtract_extend<mode>xf3_i387): Remove.
5177 (logb<mode>2): Change operand 1 predicate to general_operand.
5178 Extend operand 1 to XFmode and generate logbxf3 insn.
5179 (ilogb<mode>2): Change operand 1 predicate to general_operand.
5180 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
5181 (significand<mode>2): Change operand 1 predicate to general_operand.
5182 Extend operand 1 to XFmode and generate significandxf3 insn.
5184 2018-09-11 Nathan Sidwell <nathan@acm.org>
5186 * gcc.c (perror_with_name, pfatal_with_name): Delete.
5187 (load_specs): Use fatal_error.
5188 (DELETE_IF_ORDINARY, process_command): Use error.
5189 (execute, run_attempt): Use fatal_error.
5191 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
5193 * diagnostic-core.h (sorry_at): New prototype.
5194 * diagnostic.c (sorry_at): New function.
5196 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
5198 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
5199 by zero as VR_UNDEFINED.
5201 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
5203 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
5204 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
5205 (<sincos>mode2): New expander.
5206 (sincos_extend<mode>xf3_i387): Remove insn pattern.
5207 (sincos -> sin, cos splitters): Remove splitter patterns.
5208 (sincos<mode>3): Change operand 2 predicate to general_operand.
5209 Extend operand 2 to XFmode and generate sincosxf3 insn.
5210 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
5211 Change operand 3 predicate to const1_operand.
5212 (fptan_extend<mode>xf4_i387): Remove insn pattern.
5213 (tanxf2): Update operands in the call to fptanxf4_i387.
5214 (tan<mode>2): Change operand 1 predicate to general_operand.
5215 Extend operand 1 to XFmode and generate tanxf3 insn.
5216 (atan2xf3): Rename from *fpatanxf3_i387.
5217 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
5218 (atan2xf3): Remove expander.
5219 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
5220 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
5221 (atan<mode>2): Change operand 1 predicate to general_operand.
5222 Extend operand 1 to XFmode and generate atanxf3 insn.
5224 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
5226 * config/i386/i386.md (x87/SSE constant load splitter): Use
5227 memory_operand instead of nonimmediate_operand for input operand
5230 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
5232 * config/i386/i386.md (float partial SSE register stall splitter): Move
5233 splitter near its instruction pattern.
5234 (float_extend partial SSE register stall splitter): Ditto.
5235 (float_truncate partial SSE register stall splitter): Ditto.
5237 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
5240 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
5241 to speculation_safe_value_not_needed.
5244 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
5245 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
5246 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
5247 leaf_function_p, instead use has_hard_reg_initial_val.
5249 2018-09-09 Nathan Sidwell <nathan@acm.org>
5251 * gcc.h (pfatal_with_name): Don't declare here.
5252 * gcc.c (pfatal_with_name): Make static.
5254 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
5256 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
5259 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
5261 PR rtl-optimization/85458
5262 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
5263 priority hook to reduce the priority of EXPR.
5265 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
5267 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
5268 DImode for x87 on 32bit targets. Conditionally disable x87 modes
5269 with X87_ENABLE_FLOAT. Remove preparation code.
5270 (*float<SWI48:mode><MODEF:mode>2): Rename from
5271 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
5272 math using "enabled" attribute.
5273 (*floatdi<MODEF:mode>2_i387): Rename from
5274 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
5275 enable for 32bit targets only.
5276 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
5278 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
5279 as operand 1 predicate. Rewrite as define_insn_and_split.
5280 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
5282 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
5284 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
5285 to fallthru to FLOAT case.
5287 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
5290 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
5291 around folding of vec_sl to handle out of range shift values.
5293 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
5295 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
5296 Update callers to gen_fix_trunc<mode>_i387_fisttp
5297 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
5298 nonimmediate_operand.
5299 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
5300 and corresponding splitters.
5301 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
5302 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
5303 (fix_truncdi_i387_with_temp): Remove insn pattern
5304 and corresponding splitters.
5305 (fix_trunc<mode>_i387): Change operand 0 predicate to
5306 nonimmediate_operand.
5307 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
5308 and corresponding splitters.
5309 (*fistdi2_1): Remove.
5311 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
5312 (lrintxfdi2): Remove expander. Reimplement as define_insn.
5313 (*fist<mode>2_1): Remove.
5314 (fist<mode>2): Ditto.
5315 (fist<mode>2_with_temp): Remove insn pattern and corresponding
5317 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
5318 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
5319 (fistdi2_<rounding>): Change operand 0 predicate to
5320 nonimmediate_operand.
5321 (fistdi2_<rounding>_with_temp): Remove insn pattern
5322 and corresponding splitters.
5323 (fist<mode>2_<rounding>): Change operand 0 predicate to
5324 nonimmediate_operand.
5325 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
5326 and corresponding splitters.
5328 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
5330 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
5332 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
5335 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
5337 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
5338 early gimple folding of vec_splat().
5339 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
5340 * gimple-fold.h: Add an extern define for tree_vec_extract().
5342 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
5344 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
5345 wrappers around TREE_TYPE comparisons.
5347 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
5350 * config/s390/predicates.md: Add nonsym_memory_operand.
5351 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
5352 contains a SYMBOL_REF, load it into an intermediate pseudo.
5353 (s390_emit_compare_and_swap): Legitimize operand.
5354 * config/s390/s390.md: Use the new nonsym_memory_operand
5355 with UNSPECV_CAS patterns.
5357 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
5360 * config/s390/s390-passes.def: New file.
5361 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
5363 (make_pass_s390_early_mach): Add declaration.
5364 * config/s390/s390.c (make_pass_s390_early_mach):
5365 (s390_option_override): Remove dynamic registration.
5366 * config/s390/t-s390: Add s390-passes.def.
5368 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
5370 * config/s390/s390.c (s390_decompose_constant_pool_ref):
5371 Remove UNSPEC_LTREL_BASE check.
5372 (annotate_constant_pool_refs): Likewise.
5373 (find_constant_pool_ref): Likewise.
5374 (find_ltrel_base): Removed.
5375 (replace_ltrel_base): Removed.
5376 (s390_mainpool_finish): Remove replace_ltrel_base call.
5377 (s390_chunkify_start): Remove pending LTREL_BASE logic.
5378 (s390_chunkify_finish): Remove replace_ltrel_base call.
5379 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
5381 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
5384 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
5385 to speculation_safe_value_not_needed.
5387 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
5388 Bernd Schmidt <bernds_cb1@t-online.de>
5390 * config/nvptx/nvptx-opts.h: New file.
5391 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
5392 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
5394 (TARGET_SM35): New macro.
5395 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
5397 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
5399 (misa=): New option.
5400 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
5402 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
5404 * config/i386/i386.md (truncdfsf2): Remove expander.
5405 (truncdfsf2_with_temp): Ditto.
5406 (truncxf<mode>2): Ditto.
5407 (*truncdfsf_fast_mixed): Remove insn pattern.
5408 (*truncdfsf_fast_i387): Ditto.
5409 (*truncdfsf_mixed): Ditto.
5410 (*truncdfsf_i387): Ditto.
5411 (*truncdfsf2_i387_1): Ditto.
5412 (*truncxfsf2_mixed): Ditto.
5413 (*truncxfdf2_mixed): Ditto.
5414 (*truncxf<mode>2_i387_noop): Ditto. Update callers
5415 to call gen_truncxf<mode>2 instead.
5416 (*truncxf<mode>2_i387): Remove.
5417 (reg->reg splitters): Remove splitter pattern.
5418 (reg->mem splitters): Ditto.
5420 (truncdfsf2): New insn pattern.
5421 (truncxf<mode>2): Ditto.
5423 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5425 * tree-ssa-math-opts.c (is_mult_by): New function.
5426 (is_square_of): Use the above.
5427 (optimize_recip_sqrt): New function.
5428 (pass_cse_reciprocals::execute): Use the above.
5430 2018-09-05 Richard Biener <rguenther@suse.de>
5433 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
5434 to zero-init the emplaced vec.
5436 2018-09-05 Martin Liska <mliska@suse.cz>
5438 PR tree-optimization/87205
5439 * tree-switch-conversion.c (pass_lower_switch::execute):
5440 Group cases for switch statements.
5442 2018-09-05 Richard Biener <rguenther@suse.de>
5444 PR tree-optimization/87217
5445 * tree-ssa-sccvn.c (vuse_valueize): New.
5446 (vn_reference_lookup_pieces): Use it.
5447 (vn_reference_lookup): Likewise.
5449 2018-09-05 Nathan Sidwell <nathan@acm.org>
5452 * stor-layout.c (place_field): Scan forwards to check last
5453 bitfield when ms_bitfield_placement is in effect.
5455 2018-09-05 Richard Biener <rguenther@suse.de>
5458 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
5461 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
5462 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
5464 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
5465 * config.gcc (extra_objs): Build it.
5466 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
5468 * config/aarch64/aarch64-passes.def
5469 (pass_tag_collision_avoidance): New pass.
5470 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
5471 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
5472 (aarch64_classify_address): Remove static qualifier.
5473 (aarch64_address_info, aarch64_address_type): Move to...
5474 * config/aarch64/aarch64-protos.h: ... here.
5475 (make_pass_tag_collision_avoidance): New function.
5476 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
5479 2018-09-05 Martin Liska <mliska@suse.cz>
5481 * doc/gcov.texi: Update documentation of humar
5483 * gcov.c (format_count): Print one decimal place, it provides
5484 more fine number of situations like '1G' vs. '1.4G'.
5486 2018-09-05 Martin Liska <mliska@suse.cz>
5489 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
5490 * optc-gen.awk: Allow 'Var' for Deprecated options in order
5491 to generate a MASK value.
5493 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
5496 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
5497 if frame pointer isn't used.
5498 (compute_frame_pointer_to_fb_displacement): Likewise.
5500 2018-09-04 Jakub Jelinek <jakub@redhat.com>
5503 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
5504 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
5505 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
5506 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
5507 and OPTION_MASK_ISA_XSAVEC_UNSET.
5509 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
5511 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
5512 XOR operations in NAND case.
5514 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
5516 * wide-int-range.cc (wide_int_range_convert): New.
5517 * wide-int-range.h (wide_int_range_convert): New.
5518 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
5519 code into wide_int_range_convert.
5520 (extract_range_into_wide_ints): Do not munge anti range constants
5521 into the entire domain. Just return the range back.
5523 2018-09-04 Martin Liska <mliska@suse.cz>
5525 * genmatch.c (output_line_directive): Add new argument
5527 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
5529 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
5531 * doc/invoke.texi (Option Summary): Add whitespace.
5533 * doc/invoke.texi (Option Summary): Add -Waligned-new.
5535 2018-09-04 Richard Biener <rguenther@suse.de>
5537 PR tree-optimization/87211
5538 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
5539 backedge value we're supposed to treat as VARYING also number
5540 the PHI to VARYING in case it got a different value-number already.
5542 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
5544 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
5545 (extract_range_from_binary_expr_1): Do not call
5546 vrp_can_optimize_bit_op.
5547 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
5549 (wide_int_range_get_mask_and_bounds): New.
5550 (wide_int_range_optimize_bit_op): New.
5551 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
5552 (wide_int_range_bit_and): Same.
5553 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
5554 (wide_int_range_optimize_bit_op): New.
5555 (wide_int_range_get_mask_and_bounds): New.
5557 2018-09-04 Richard Biener <rguenther@suse.de>
5559 PR tree-optimization/87176
5560 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
5561 variable. When value-numbering a virtual PHI node make sure
5562 to not value-number to the backedge value.
5564 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
5566 * doc/extend.texi (Long Long, Hex Floats): Document support for
5567 long long and hex floats in more recent versions of ISO C++.
5569 2018-09-03 Richard Biener <rguenther@suse.de>
5571 PR tree-optimization/87177
5572 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
5575 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
5577 * bb-reorder.c (edge_order): Convert to C-qsort-style
5578 tri-state comparator.
5579 (reorder_basic_blocks_simple): Change std::stable_sort to
5582 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
5584 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
5585 tri-state comparator.
5586 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
5588 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
5590 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
5591 (mergesort): ... here as maximum count for using netsort.
5592 (gcc_qsort): Set nlim to 3 if stable sort is requested.
5593 (gcc_stablesort): New.
5594 * system.h (gcc_stablesort): Declare.
5596 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
5598 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
5599 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
5600 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
5602 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
5604 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
5605 lxsdx and stxsdx alternatives.
5606 (*mov<mode>_hardfloat64): Ditto.
5607 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
5609 2018-09-03 Richard Biener <rguenther@suse.de>
5611 PR tree-optimization/87200
5612 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
5615 2018-09-03 Martin Liska <mliska@suse.cz>
5617 PR tree-optimization/87201
5618 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
5619 Fix parenthesis in an expression.
5621 2018-09-03 Richard Biener <rguenther@suse.de>
5623 PR tree-optimization/87197
5624 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
5625 visited. CSE the VN_INFO hashtable lookup.
5627 PR tree-optimization/87169
5628 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
5629 iterating make sure there's no extra backedges from irreducible
5630 regions feeding the header. Mark the destination block
5633 2018-09-03 Martin Liska <mliska@suse.cz>
5636 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
5637 * common/common-targhooks.c (default_get_valid_option_values):
5639 * common/common-targhooks.h (default_get_valid_option_values):
5641 * common/config/i386/i386-common.c: Move processor_target_table
5643 (ix86_get_valid_option_values): New function.
5644 (TARGET_GET_VALID_OPTION_VALUES): New macro.
5645 * config/i386/i386.c (struct ptt): Move to i386-common.c.
5646 (PTA_*): Move all defined masks into i386-common.c.
5647 (ix86_function_specific_restore): Use new processor_cost_table.
5648 * config/i386/i386.h (struct ptt): Moved from i386.c.
5649 (struct pta): Likewise.
5650 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
5651 * doc/tm.texi.in: Likewise.
5652 * opt-suggestions.c (option_proposer::suggest_option):
5653 Pass prefix to build_option_suggestions.
5654 (option_proposer::get_completions): Likewise.
5655 (option_proposer::build_option_suggestions): Use the new target
5657 * opts.c (struct option_help_tuple): New struct.
5658 (print_filtered_help): Use the new target hook.
5660 2018-09-03 Martin Liska <mliska@suse.cz>
5663 * predict.c (set_even_probabilities): Add likely_edges
5664 argument and handle cases where we have precisely one
5666 (combine_predictions_for_bb): Catch also likely_edges.
5667 (tree_predict_by_opcode): Handle gswitch statements.
5668 * tree-cfg.h (find_case_label_for_value): New declaration.
5669 (find_taken_edge_switch_expr): Likewise.
5670 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
5671 Find pivot in decision tree based on probabily, not by number of
5674 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
5676 * doc/standards.texi (Standards): Update Objective-C reference.
5678 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
5680 * doc/install.texi (Prerequisites): Update link for MPC.
5682 2018-09-01 Michael Matz <matz@suse.de>
5684 PR tree-optimization/87074
5685 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
5686 PHIs for outer-loop uses.
5688 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
5690 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
5691 * doc/invoke.texi (C Dialect Options): Ditto.
5693 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
5695 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
5697 2018-08-31 Richard Biener <rguenther@suse.de>
5699 PR tree-optimization/87168
5700 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
5701 (rpo_elim::eliminate_avail): When OP was not visited it must
5704 2018-08-31 David Malcolm <dmalcolm@redhat.com>
5706 * tree-vrp.c (copy_value_range): Convert param "from" from
5707 "value_range *" to "const value_range *".
5708 (range_is_null): Likewise for param "vr".
5709 (range_int_cst_p): Likewise.
5710 (range_int_cst_singleton_p): Likewise.
5711 (symbolic_range_p): Likewise.
5712 (value_ranges_intersect_p): Likewise for both params.
5713 (value_range_nonnegative_p): Likewise for param "vr".
5714 (value_range_constant_singleton): Likewise.
5715 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
5716 (extract_range_into_wide_ints): Likewise for param "vr".
5717 (extract_range_from_multiplicative_op): Likewise for params "vr0"
5719 (vrp_can_optimize_bit_op): Likewise.
5720 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
5722 (extract_range_from_unary_expr): Likewise.
5723 (debug_value_range): Likewise for param "vr".
5724 (value_range::dump): Add "const" qualifier.
5725 (vrp_prop::check_array_ref): Convert local "vr" from
5726 "value_range *" to "const value_range *".
5727 (vrp_prop::check_mem_ref): Likewise.
5728 (vrp_prop::visit_stmt): Likewise for local "old_vr".
5729 (vrp_intersect_ranges_1): Likewise for param "vr_1".
5730 (vrp_intersect_ranges): Likewise.
5731 (simplify_stmt_for_jump_threading): Likewise for local "vr".
5732 (vrp_prop::vrp_finalize): Likewise.
5733 * tree-vrp.h (value_range::dump): Add "const" qualifier.
5734 (vrp_intersect_ranges): Add "const" qualifier to params as above.
5735 (extract_range_from_unary_expr): Likewise.
5736 (value_range_constant_singleton): Likewise.
5737 (symbolic_range_p): Likewise.
5738 (copy_value_range): Likewise.
5739 (extract_range_from_binary_expr_1): Likewise.
5740 (range_int_cst_p): Likewise.
5741 (vrp_set_zero_nonzero_bits): Likewise.
5742 (range_int_cst_singleton_p): Likewise.
5744 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
5746 * config/aarch64/arm_neon.h (vabsd_s64): New.
5747 (vnegd_s64): Likewise.
5749 2018-08-31 Martin Jambor <mjambor@suse.cz>
5751 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
5753 2018-08-31 Martin Liska <mliska@suse.cz>
5755 * ipa-icf.c (sem_item::add_type): Use
5756 sem_item::m_type_hash_cache.
5757 * ipa-icf.h: Move the cache from sem_item_optimizer
5760 2018-08-31 Nathan Sidwell <nathan@acm.org>
5762 * doc/extend.texi (Backwards Compatibility): Remove implicit
5763 extern C leeway of () being (...).
5765 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5767 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
5769 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
5773 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
5775 2018-08-31 Jakub Jelinek <jakub@redhat.com>
5778 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
5779 gen_int_mode. Formatting fixes.
5781 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
5783 * target.def (custom_function_descriptors): Improve documentation.
5784 * doc/tm.texi.in (Trampolines): Expand discussion of function
5785 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
5786 beginning of the section.
5787 * doc/tm.texi: Regenerated.
5789 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
5791 * cfg.h (class auto_edge_flag): Spell out the template-id of the
5792 base class in the initializer list. This is a workaround for
5793 building with older GCC.
5794 (class auto_bb_flag): Likewise.
5796 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
5798 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
5799 (altivec_vcmpequ<VI_char>_p): Remove star.
5800 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
5802 (expand_strncmp_vec_sequence): New function.
5803 (emit_final_str_compare_vec): New function.
5804 (expand_strn_compare): Add support for vector strncmp.
5805 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
5806 length specification to bytes.
5807 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
5808 (vcmpnezb_p): New pattern.
5809 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
5810 for option -mstring-compare-inline-limit.
5812 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
5814 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
5815 (PTA_SKYLAKE): Add PTA_AES.
5816 (PTA_GOLDMONT): Likewise.
5818 2018-08-29 Jan Hubicka <jh@suse.cz>
5821 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
5822 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
5824 2018-08-29 Jan Hubicka <jh@suse.cz>
5826 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
5828 (hash_tree): Do not visit TYPE_STUB_DECL.
5829 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
5830 stream TYPE_STUB_DECL.
5831 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
5832 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
5833 after free_lang_data.
5834 (type_in_anonymous_namespace_p): Likewise.
5836 2018-08-29 Jan Hubicka <jh@suse.cz>
5838 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
5839 comment that it has to be even number.
5840 (class sreal): Change m_sig type to int32_t.
5841 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
5842 int64_t for temporary calculations.
5843 (sreal_verify_basics): Drop one bit from minimum and maximum.
5845 2018-08-30 Richard Biener <rguenther@suse.de>
5847 PR tree-optimization/87147
5848 * tree-ssa-sccvn.c (SSA_VISITED): New function.
5849 (visit_phi): When the degenerate result is from the backedge and
5850 we didn't visit its definition yet drop to VARYING.
5851 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
5853 2018-08-29 Jan Hubicka <jh@suse.cz>
5855 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
5857 (hash_tree): Likewise.
5859 2018-08-29 Jan Hubicka <jh@suse.cz>
5861 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
5862 and TYPE_NEXT_REF_TO.
5864 2018-08-29 Jan Hubicka <jh@suse.cz>
5866 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
5867 comment that it has to be even number.
5868 (class sreal): Change m_sig type to int32_t.
5869 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
5870 int64_t for temporary calculations.
5871 (sreal_verify_basics): Drop one bit from minimum and maximum.
5873 2018-08-30 Tamar Christina <tamar.christina@arm.com>
5875 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
5877 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
5880 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
5881 if to_add is negative.
5883 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
5886 * builtins.c (c_strlen): Improve range checks.
5888 2018-08-29 Martin Sebor <msebor@redhat.com>
5889 Jeff Law <law@redhat.com>
5891 PR tree-optimization/86714
5892 PR tree-optimization/86711
5893 * builtins.c (c_strlen): Add arguments to call to string_constant.
5894 * expr.c (string_constant): Add argument. Detect missing nul
5895 terminator and outermost declaration it's missing in.
5896 * expr.h (string_constant): Add argument.
5897 * fold-const.c (read_from_constant_string): Add arguments to call to
5899 (c_getstr): Likewise.
5900 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
5902 * tree-ssa-strlen.c (get_stridx): Likewise.
5904 2018-08-29 Jan Hubicka <jh@suse.cz>
5906 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
5907 Do not stream DECL_VINDEX.
5908 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
5909 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
5910 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
5912 2018-08-29 Richard Biener <rguenther@suse.de>
5914 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
5915 virtual operands that are not default defs to honor region
5917 (rpo_vn_valueize): Remove ineffective code here.
5919 2018-08-29 Richard Biener <rguenther@suse.de>
5921 PR tree-optimization/87132
5922 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
5923 when skipping defs reachable over backedges.
5925 2018-08-29 Richard Biener <rguenther@suse.de>
5927 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
5928 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
5929 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
5930 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
5931 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
5932 (vn_reference_lookup_3): Remove use of const_parms.
5933 (free_rpo_vn): Do not free const_parms.
5934 (do_rpo_vn): Do not call init_const_parms.
5935 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
5936 SSA_NAME_POINTS_TO_READONLY_MEMORY.
5937 (call_may_clobber_ref_p_1): Likewise.
5939 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
5942 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
5944 (-ftree-scev-cprop): Document.
5946 2018-08-29 Jan Hubicka <jh@suse.cz>
5948 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
5950 (sreal constructor): Update.
5951 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
5952 sreal:operator/): Update.
5954 2018-08-29 Martin Liska <mliska@suse.cz>
5956 * tree-switch-conversion.c (switch_conversion::expand):
5957 Strenghten assumption about gswitch statements.
5959 2018-08-29 Richard Biener <rguenther@suse.de>
5961 PR tree-optimization/87117
5962 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
5963 re-value-number released SSA VDEFs.
5965 2018-08-29 Richard Biener <rguenther@suse.de>
5967 PR tree-optimization/87126
5968 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
5970 2018-08-28 Jim Wilson <jimw@sifive.com>
5972 * config/riscv/pic.md: Rewrite.
5973 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
5975 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
5976 (SOFTF, default_load, softload, softstore): New.
5978 2018-08-28 Jeff Law <law@redhat.com>
5980 * fold-const.c (fold_binary_loc): Remove recently added assert.
5982 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
5984 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
5985 to OP parmeter of generated function.
5987 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
5989 PR tree-optimization/87009
5990 * match.pd: Add boolean optimizations.
5992 2018-08-28 Martin Sebor <msebor@redhat.com>
5995 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
5996 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
5997 (pass_walloca::gate): Use it.
5998 (alloca_call_type): Same.
5999 (pass_walloca::execute): Same.
6000 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
6002 2018-08-28 David Malcolm <dmalcolm@redhat.com>
6004 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
6005 GCC_VERSION for usage of "__gcc_dump_printf__" format from
6008 2018-08-28 Richard Biener <rguenther@suse.de>
6010 PR tree-optimization/87124
6011 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
6012 constants before looking up avail.
6014 2018-08-28 Jakub Jelinek <jakub@redhat.com>
6017 * calls.c (maybe_warn_nonstring_arg): Punt early if
6018 warn_stringop_overflow is zero. Don't call get_range_strlen
6019 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
6020 Swap comparison operands to have constants on rhs. Only use
6021 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
6022 increment lenrng[0].
6024 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
6026 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
6027 use of tree_to_shwi. Remove duplicated test for the size being
6028 a whole number of bytes.
6030 2018-08-28 Richard Biener <rguenther@suse.de>
6032 PR tree-optimization/87117
6033 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
6034 Handle removed stmt without LHS (GIMPLE_NOP).
6036 2018-08-28 Richard Biener <rguenther@suse.de>
6038 PR tree-optimization/87117
6039 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
6040 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
6042 2018-08-28 Richard Biener <rguenther@suse.de>
6044 PR tree-optimization/87117
6045 * tree-ssa-pre.c (compute_avail): Do not make expressions
6046 with predicated values available.
6047 (get_expr_value_id): Assert we do not run into predicated value
6050 2018-08-28 Richard Biener <rguenther@suse.de>
6052 PR tree-optimization/87117
6053 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
6054 get virtual operands.
6055 (get_expr_operands): Handle STRING_CST like other decls.
6057 2018-08-28 Martin Liska <mliska@suse.cz>
6059 * tree.h: Update documentation of fndecl_built_in_p
6063 2018-08-27 Jeff Law <law@redhat.com>
6064 PR tree-optimization/87110
6065 * tree-ssa-dse.c (compute_trims): Handle non-constant
6068 2018-08-27 Martin Sebor <msebor@redhat.com>
6070 PR tree-optimization/86914
6071 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
6073 2018-08-27 Martin Sebor <msebor@redhat.com>
6075 PR tree-optimization/87112
6076 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
6077 the type of the bound argument.
6079 2018-08-27 Jeff Law <law@redhat.com>
6081 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
6082 type does not have a TYPE_SIZE_UNIT.
6084 2018-08-27 Steve Ellcey <sellcey@cavium.com>
6086 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
6087 with include of backend.h.
6089 2018-08-27 Richard Biener <rguenther@suse.de>
6091 PR tree-optimization/86927
6092 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
6093 use const cond reduction code.
6095 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
6097 PR tree-optimization/85758
6098 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
6100 2018-08-27 David Malcolm <dmalcolm@redhat.com>
6103 * diagnostic-show-locus.c (class layout_range): Update for
6104 conversion of show_caret_p to a tri-state.
6105 (layout_range::layout_range): Likewise.
6106 (make_range): Likewise.
6107 (layout::maybe_add_location_range): Likewise.
6108 (layout::should_print_annotation_line_p): Don't show annotation
6109 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
6110 (layout::get_state_at_point): Update for conversion of
6111 show_caret_p to a tri-state. Bail out early for
6112 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
6113 underlining or source colorization.
6114 (gcc_rich_location::add_location_if_nearby): Update for conversion
6115 of show_caret_p to a tri-state.
6116 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
6117 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
6119 (selftest::test_one_liner_labels): Likewise.
6120 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
6121 conversion of show_caret_p to a tri-state.
6122 * pretty-print.c (text_info::set_location): Likewise.
6123 * pretty-print.h (text_info::set_location): Likewise.
6124 * substring-locations.c (format_warning_n_va): Likewise.
6125 * tree-diagnostic.c (default_tree_printer): Likewise.
6126 * tree-pretty-print.c (newline_and_indent): Likewise.
6128 2018-08-27 David Malcolm <dmalcolm@redhat.com>
6131 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
6132 line above for line-insertion fix-it hints.
6133 (selftest::test_fixit_insert_containing_newline): Update the
6134 expected results, and add a test with line-numbering enabled.
6136 2018-08-27 Martin Liska <mliska@suse.cz>
6139 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
6140 params with DECL_HAS_VALUE_EXPR_P.
6142 2018-08-27 Martin Liska <mliska@suse.cz>
6144 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
6145 selected expansion strategy.
6147 2018-08-27 Martin Liska <mliska@suse.cz>
6149 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
6150 * builtins.c (is_builtin_fn): Likewise.
6151 * attribs.c (diag_attr_exclusions): Use new function
6152 fndecl_built_in_p and remove check for FUNCTION_DECL if
6154 (builtin_mathfn_code): Likewise.
6155 (fold_builtin_expect): Likewise.
6156 (fold_call_expr): Likewise.
6157 (fold_builtin_call_array): Likewise.
6158 (fold_call_stmt): Likewise.
6159 (set_builtin_user_assembler_name): Likewise.
6160 (is_simple_builtin): Likewise.
6161 * calls.c (gimple_alloca_call_p): Likewise.
6162 (maybe_warn_nonstring_arg): Likewise.
6163 * cfgexpand.c (expand_call_stmt): Likewise.
6164 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
6165 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
6166 (cgraph_node::verify_node): Likewise.
6167 * cgraphclones.c (build_function_decl_skip_args): Likewise.
6168 (cgraph_node::create_clone): Likewise.
6169 * config/arm/arm.c (arm_insert_attributes): Likewise.
6170 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
6171 * dse.c (scan_insn): Likewise.
6172 * expr.c (expand_expr_real_1): Likewise.
6173 * fold-const.c (operand_equal_p): Likewise.
6174 (fold_binary_loc): Likewise.
6175 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
6176 * gimple-low.c (lower_stmt): Likewise.
6177 * gimple-pretty-print.c (dump_gimple_call): Likewise.
6178 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
6180 * gimple.c (gimple_build_call_from_tree): Likewise.
6181 (gimple_call_builtin_p): Likewise.
6182 (gimple_call_combined_fn): Likewise.
6183 * gimplify.c (gimplify_call_expr): Likewise.
6184 (gimple_boolify): Likewise.
6185 (gimplify_modify_expr): Likewise.
6186 (gimplify_addr_expr): Likewise.
6187 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
6188 * ipa-cp.c (determine_versionability): Likewise.
6189 * ipa-fnsummary.c (compute_fn_summary): Likewise.
6190 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
6191 * ipa-split.c (visit_bb): Likewise.
6192 (split_function): Likewise.
6193 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
6194 * lto-cgraph.c (input_node): Likewise.
6195 * lto-streamer-out.c (write_symbol): Likewise.
6196 * omp-low.c (setjmp_or_longjmp_p): Likewise.
6197 (lower_omp_1): Likewise.
6198 * predict.c (strip_predict_hints): Likewise.
6199 * print-tree.c (print_node): Likewise.
6200 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
6201 * trans-mem.c (is_tm_irrevocable): Likewise.
6202 (is_tm_load): Likewise.
6203 (is_tm_simple_load): Likewise.
6204 (is_tm_store): Likewise.
6205 (is_tm_simple_store): Likewise.
6206 (is_tm_abort): Likewise.
6207 (tm_region_init_1): Likewise.
6208 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
6209 * tree-cfg.c (verify_gimple_call): Likewise.
6210 (move_stmt_r): Likewise.
6211 (stmt_can_terminate_bb_p): Likewise.
6212 * tree-eh.c (lower_eh_constructs_2): Likewise.
6213 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
6214 * tree-inline.c (remap_gimple_stmt): Likewise.
6215 (copy_bb): Likewise.
6216 (estimate_num_insns): Likewise.
6217 (fold_marked_statements): Likewise.
6218 * tree-sra.c (scan_function): Likewise.
6219 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
6220 (optimize_stack_restore): Likewise.
6221 (pass_fold_builtins::execute): Likewise.
6222 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
6223 (mark_all_reaching_defs_necessary_1): Likewise.
6224 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
6225 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
6226 (pass_forwprop::execute): Likewise.
6227 * tree-ssa-loop-im.c (stmt_cost): Likewise.
6228 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
6229 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
6230 * tree-ssa-strlen.c (get_string_length): Likewise.
6231 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
6232 (find_func_aliases_for_call): Likewise.
6233 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
6234 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
6235 * tree-tailcall.c (find_tail_calls): Likewise.
6236 * tree.c (need_assembler_name_p): Likewise.
6237 (free_lang_data_in_decl): Likewise.
6238 (get_call_combined_fn): Likewise.
6239 * ubsan.c (is_ubsan_builtin_p): Likewise.
6240 * varasm.c (incorporeal_function_p): Likewise.
6241 * tree.h (DECL_BUILT_IN): Remove and replace with
6243 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
6244 (fndecl_built_in_p): New.
6246 2018-08-27 Martin Liska <mliska@suse.cz>
6248 PR tree-optimization/86847
6249 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
6250 Dump also subtree probability.
6251 (switch_decision_tree::do_jump_if_equal): New function.
6252 (switch_decision_tree::emit_case_nodes): Handle special
6253 situations in balanced tree that can be emitted much simpler.
6254 Fix calculation of probabilities that happen in tree expansion.
6255 * tree-switch-conversion.h (struct cluster): Add
6257 (struct simple_cluster): Likewise.
6258 (struct case_tree_node): Add new function has_child.
6259 (do_jump_if_equal): New.
6261 2018-08-27 Martin Liska <mliska@suse.cz>
6263 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
6264 Add new argument to bit_test_cluster constructor.
6265 (bit_test_cluster::emit): Set bits really number of values
6267 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
6268 probability argument.
6269 * tree-switch-conversion.h (struct bit_test_cluster):
6270 Add m_handles_entire_switch.
6272 2018-08-27 Martin Liska <mliska@suse.cz>
6274 PR tree-optimization/86702
6275 * tree-switch-conversion.c (jump_table_cluster::emit):
6276 Make probabilities even for values in jump table
6277 according to number of cases handled.
6278 (switch_decision_tree::compute_cases_per_edge): Pass
6279 argument to reset_out_edges_aux function.
6280 (switch_decision_tree::analyze_switch_statement): Likewise.
6281 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
6284 2018-08-27 Martin Liska <mliska@suse.cz>
6286 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
6287 cfun argument explicitly.
6288 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
6289 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
6290 function gimple_switch_default_bb.
6291 (convert_switch_statements):
6293 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
6294 * stmt.c (label_to_block_fn): Use label_to_block and pass
6295 cfun argument explicitly and use gimple_switch_label_bb.
6296 (expand_case): Likewise.
6297 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
6298 cfun argument explicitly. Likewise.
6299 (make_edges_bb): Likewise.
6300 (make_cond_expr_edges): Likewise.
6301 (get_cases_for_edge): Likewise.
6302 (make_gimple_switch_edges): Likewise.
6303 (label_to_block_fn): Likewise.
6304 (label_to_block): Likewise.
6305 (make_goto_expr_edges): Likewise.
6306 (make_gimple_asm_edges): Likewise.
6307 (main_block_label): Likewise.
6308 (group_case_labels_stmt): Likewise.
6309 (find_taken_edge_computed_goto): Likewise.
6310 (find_taken_edge_switch_expr): Likewise.
6311 (gimple_verify_flow_info): Likewise.
6312 (gimple_redirect_edge_and_branch): Likewise.
6313 (gimple_switch_label_bb): New function.
6314 (gimple_switch_default_bb): Likewise.
6315 (gimple_switch_edge): Likewise.
6316 (gimple_switch_default_edge): Likewise.
6317 * tree-cfg.h (label_to_block_fn): Remove and replace ...
6318 (label_to_block): ... with this.
6319 (gimple_switch_label_bb): New.
6320 (gimple_switch_default_bb): Likewise.
6321 (gimple_switch_edge): Likewise.
6322 (gimple_switch_default_edge): Likewise.
6323 * tree-cfgcleanup.c (convert_single_case_switch): Use
6324 new gimple functions and pass new argument to label_to_block.
6325 (cleanup_control_flow_bb):
6326 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
6327 cfun argument explicitly.
6328 (make_eh_edges): Likewise.
6329 (redirect_eh_dispatch_edge): Likewise.
6330 (lower_resx): Likewise.
6331 (lower_eh_dispatch): Likewise.
6332 (maybe_remove_unreachable_handlers): Likewise.
6333 (unsplit_eh): Likewise.
6334 (cleanup_empty_eh): Likewise.
6335 (verify_eh_edges): Likewise.
6336 (verify_eh_dispatch_edge): Likewise.
6337 * tree-ssa-dom.c (record_edge_info): Likewise.
6338 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
6339 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
6340 (thread_through_normal_block): Likewise.
6341 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
6342 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
6343 * tree-switch-conversion.c (switch_conversion::collect): Use new
6345 (switch_conversion::check_final_bb): Likewise.
6346 (switch_conversion::gather_default_values): Pass new argument
6348 (switch_conversion::build_constructors): Likewise.
6349 (switch_decision_tree::compute_cases_per_edge): Use new
6350 gimple_switch_edge function.
6351 (switch_decision_tree::analyze_switch_statement): Pass new argument
6353 (switch_decision_tree::try_switch_expansion): Use
6354 gimple_switch_default_edge.
6355 * tree-vrp.c (find_switch_asserts): Pass new argument
6357 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
6358 (vr_values::simplify_switch_using_ranges): Likewise.
6360 2018-08-27 Richard Biener <rguenther@suse.de>
6362 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
6363 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
6365 * tree-ssa-sccvn.h (struct vn_pval): New structure.
6366 (struct vn_nary_op_s): Add unwind_to member. Add
6367 predicated_values flag and put result into a union together
6368 with a linked list of vn_pval.
6369 (struct vn_ssa_aux): Add name member to make maintaining
6370 a map of SSA name to vn_ssa_aux possible. Remove no longer
6371 needed info, dfsnum, low, visited, on_sccstack, use_processed
6372 and range_info_anti_range_p members.
6373 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
6374 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
6376 (vn_valueize): New global.
6377 (vn_context_bb): Likewise.
6378 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
6379 VN_INFO_PTR_INFO): Remove.
6380 * tree-ssa-sccvn.c: ... (rewrite)
6381 (pass_fre::execute): For -O2+ initialize loops and run
6382 RPO VN in optimistic mode (iterating). For -O1 and -Og
6383 run RPO VN in non-optimistic mode.
6384 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
6385 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
6386 * doc/invoke.texi (sccvn-max-scc-size): Remove.
6387 (rpo-vn-max-loop-depth): Document.
6388 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
6389 when valuezing the VUSE signals we walked out of the region.
6390 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
6391 (phi_translate): Set VN context block to use for availability
6393 (compute_avail): Likewise.
6394 (pre_valueize): New function.
6395 (pass_pre::execute): Adjust to the RPO VN API.
6397 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
6398 (propagate_constants_for_unrolling): Remove.
6399 (tree_unroll_loops_completely): Perform value-numbering
6400 on the unrolled bodies loop parent.
6402 2018-08-27 Richard Biener <rguenther@suse.de>
6404 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
6405 for partial antic compute.
6407 2018-08-27 Jakub Jelinek <jakub@redhat.com>
6409 PR rtl-optimization/87065
6410 * combine.c (simplify_if_then_else): Formatting fix.
6411 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
6413 (known_cond): Don't return const_true_rtx for vector modes. Use
6414 CONST0_RTX instead of const0_rtx. Formatting fixes.
6416 2018-08-27 Martin Liska <mliska@suse.cz>
6418 PR gcov-profile/87069
6419 * gcov.c (process_file): Record files already processed
6420 and warn about a file being processed multiple times.
6422 2018-08-27 Martin Liska <mliska@suse.cz>
6425 * config/aarch64/aarch64.c (aarch64_override_options_internal):
6426 Set default values for x_aarch64_*_string strings.
6427 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
6428 prefix. For -mabi do not print '=ABI' in help and use
6429 <option_value> format for -msve-vector-bits and -moverride
6432 2018-08-26 Jeff Law <law@redhat.com>
6434 * config/mips/frame-header-opt.c: Include "backend.h" rather than
6437 2018-08-26 Marek Polacek <polacek@redhat.com>
6439 PR c++/87029, Implement -Wredundant-move.
6440 * doc/invoke.texi: Document -Wredundant-move.
6442 2018-08-25 Martin Sebor <msebor@redhat.com>
6444 PR tree-optimization/87059
6445 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
6446 to the same type as the other.
6447 * fold-const.c (fold_binary_loc): Assert expectation.
6449 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
6451 * config/darwin.c (machopic_legitimize_pic_address): Clean up
6452 extraneous parentheses, dead code section and formatting.
6454 2018-08-24 David Malcolm <dmalcolm@redhat.com>
6457 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
6458 wide enough for jumps in the line-numbering to be visible.
6459 (layout::print_gap_in_line_numbering): New member function.
6460 (layout::calculate_line_spans): When using line numbering, merge
6461 line spans that are only 1 line apart.
6462 (diagnostic_show_locus): When printing line numbers, show gaps in
6463 line numbering directly, rather than printing headers.
6464 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
6465 line-numbering with multiple line spans.
6466 (selftest::test_fixit_insert_containing_newline_2): Add test of
6467 line-numbering, in which the spans are close enough to be merged.
6469 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
6471 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
6472 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
6473 * tree-vrp.c (range_is_nonnull): Remove.
6474 (range_includes_zero_p): Accept value_range instead of min/max.
6475 (extract_range_from_binary_expr_1): Do not early bail on
6477 Use range_includes_zero_p instead of range_is_nonnull.
6478 (extract_range_from_unary_expr): Use range_includes_zero_p instead
6479 of range_is_nonnull.
6480 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
6481 special case VR_ANTI_RANGE.
6482 (vrp_finalize): Same.
6483 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
6485 (range_is_nonnull): Remove.
6486 * vr-values.c (vrp_stmt_computes_nonzero): Use
6487 range_includes_zero_p instead of range_is_nonnull.
6488 (extract_range_basic): Pass value_range to range_includes_zero_p
6489 instead of range_is_nonnull.
6491 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
6493 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
6494 * emit-rtl.h (rtl_data): Remove return_bnd.
6495 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
6496 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
6497 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
6498 (POINTER_BOUNDS_MODE): Remove definition.
6499 (make_pointer_bounds_mode): Remove.
6500 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
6501 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
6502 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
6503 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
6504 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
6505 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
6506 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
6508 * config/i386/i386-modes.def (BND32, BND64): Remove.
6509 * config/i386/i386.c (dbx_register_map): Remove bound registers.
6510 (dbx64_register_map): Ditto.
6511 (svr4_dbx_register_map): Ditto.
6512 (indirect_thunk_bnd_needed): Remove.
6513 (indirect_thunks_bnd_used): Ditto.
6514 (indirect_return_bnd_needed): Ditto.
6515 (indirect_return_via_cx_bnd): Ditto.
6516 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
6517 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
6518 (output_indirect_thunk): Ditto. Remove need_prefix argument.
6519 (output_indirect_thunk_function): Remove handling of
6520 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
6521 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
6522 (ix86_save_reg): Remove handling of crtl->return_bnd.
6523 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
6524 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
6525 and UNSPEC_BNDLX_ADDR.
6526 (ix86_output_indirect_branch_via_reg): Remove handling of
6527 indirect_thunk_prefix_bnd.
6528 (ix86_output_indirect_branch_via_push): Ditto.
6529 (ix86_output_function_return): Ditto.
6530 (ix86_output_indirect_function_return): Ditto.
6531 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
6532 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
6533 (CALL_USED_REGISTERS): Ditto.
6534 (REG_ALLOC_ORDER): Update for removal of bound registers.
6535 (HI_REGISTER_NAMES): Ditto.
6536 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
6537 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
6538 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
6539 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
6540 (FIRST_PSEUDO_REG): Update.
6541 (BND): Remove mode iterator.
6542 * config/i386/predicates.md (bnd_mem_operator): Remove.
6544 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
6546 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
6549 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
6551 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
6552 the case in which the permute needs only a single element and
6553 repeats for every vector of the result. Extend that case to
6554 handle variable-length vectors.
6555 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
6557 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
6560 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
6561 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
6563 2018-08-24 Richard Biener <rguenther@suse.de>
6565 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
6566 bb_flags_allocated members.
6567 (auto_flag): New RAII class for allocating flags.
6568 (auto_edge_flag): New RAII class for allocating edge flags.
6569 (auto_bb_flag): New RAII class for allocating bb flags.
6570 * cfgloop.c (verify_loop_structure): Allocate temporary edge
6572 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
6573 in favor of temporarily allocated BB flag.
6574 * hsa-brig.c: Re-order includes.
6575 * hsa-dump.c: Likewise.
6576 * hsa-regalloc.c: Likewise.
6577 * print-rtl.c: Likewise.
6578 * profile-count.c: Likewise.
6580 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
6583 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
6586 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
6589 * wide-int-range.cc (wide_int_range_div): Do not ignore result
6590 from wide_int_range_multiplicative_op.
6592 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6594 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
6597 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
6599 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
6602 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
6604 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
6605 full GENERIC expression used for replacement.
6607 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
6609 * tree-vrp.c (abs_extent_range): Remove.
6610 (extract_range_into_wide_ints): Pass wide ints by reference.
6611 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
6612 Pass wide ints by reference in all calls to
6613 extract_range_into_wide_ints.
6614 * wide-int-range.cc (wide_int_range_div): New.
6615 * wide-int-range.h (wide_int_range_div): New.
6616 (wide_int_range_includes_zero_p): New.
6617 (wide_int_range_zero_p): New.
6619 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
6621 * config/aarch64/aarch64.md (arches): New enum.
6622 (arch): New enum attr.
6623 (arch_enabled): New attr.
6624 (enabled): Now uses arch_enabled only.
6625 (simd, sve, fp16): Removed attribute.
6626 (fp): Attr now defined in terms of 'arch'.
6627 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
6628 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
6629 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
6630 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
6631 attributes into 'arch'.
6632 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
6633 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
6634 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
6635 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
6636 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
6637 'simd' attribute into 'arch'.
6638 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
6639 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
6640 Convert use of 'fp' attribute to 'arch'.
6641 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
6642 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
6644 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
6645 (different modes) Merge 'fp' and 'simd' into 'arch'.
6646 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
6649 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
6651 PR rtl-optimization/87026
6652 * expmed.c (canonicalize_comparison): If we can no longer create
6653 pseudoregisters, don't.
6655 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
6658 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
6660 * config/arm/arm.c (speculation_barrier_libfunc): New static
6662 (arm_init_libfuncs): Initialize it.
6663 (arm_emit_speculation_barrier): New function.
6664 * config/arm/arm.md (speculation_barrier): Call
6665 arm_emit_speculation_barrier for architectures that do not have
6667 (speculation_barrier_insn): Only match on Armv7 or later.
6669 2018-08-23 Richard Biener <rguenther@suse.de>
6672 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
6675 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
6677 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
6678 of single-vector TBLs.
6679 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
6682 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
6685 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
6686 aarch64_evpc_tbl guard.
6688 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
6690 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
6693 2018-08-22 Martin Sebor <msebor@redhat.com>
6696 * tree-pretty-print.c (pretty_print_string): Add argument.
6697 (dump_generic_node): Call to pretty_print_string with string size.
6699 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
6701 PR rtl-optimization/86771
6702 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
6703 of two SETs into those two SETs, one to be placed at i2, if that SETs
6704 destination is modified between i2 and i3.
6706 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
6708 PR tree-optimization/86725
6709 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
6711 (vect_analyze_scalar_cycles_1): Check it.
6713 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
6715 PR tree-optimization/86725
6716 * tree-vect-loop.c (vect_is_simple_reduction): When treating
6717 an outer loop phi as a double reduction, make sure that the
6718 single user of the phi result is an inner loop phi.
6720 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
6722 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
6723 grouped stores with gaps to a strided group.
6725 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
6727 * tree-vect-stmts.c (get_group_load_store_type)
6728 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
6729 first statement in a group.
6731 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
6733 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
6734 the sequence used in gcc/gcc.c.
6736 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
6739 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
6742 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
6744 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
6745 Darwin10-specific unwinder-shim.
6746 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
6747 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
6748 New to cater for Darwin10 Rosetta.
6750 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
6752 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
6753 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
6755 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
6760 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
6761 (dwarf2out_switch_text_section): Generate a local label for the
6762 second function sub-section and apply it as the second FDE start
6764 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
6765 second sub-section start.
6767 2018-08-22 Richard Biener <rguenther@suse.de>
6769 PR tree-optimization/86988
6770 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
6772 2018-08-22 Richard Biener <rguenther@suse.de>
6774 PR tree-optimization/86945
6775 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
6777 2018-08-22 Alexandre Oliva <oliva@adacore.com>
6779 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
6780 a comment about how uses of r2 for .sdata2 come about.
6782 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
6784 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
6786 2018-08-21 Marek Polacek <polacek@redhat.com>
6788 PR c++/86981, Implement -Wpessimizing-move.
6789 * doc/invoke.texi: Document -Wpessimizing-move.
6791 2018-08-21 Jan Hubicka <jh@suse.cz>
6793 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
6794 * tree.h (is_redundant_typedef): Remove.
6795 * dwarf2out.c (is_redundant_typedef): Turn into static function.
6797 2018-08-21 Jan Hubicka <jh@suse.cz>
6799 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
6802 2018-08-21 Tamar Christina <tamar.christina@arm.com>
6804 * expmed.c (extract_low_bits): Reject invalid subregs early.
6806 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
6809 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
6812 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
6814 * config/vxworks.h: Guard vxworks_asm_out_constructor and
6815 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
6816 * config/vxworks.c: Likewise.
6818 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
6820 * config/vxworks.c: Set targetm.have_ctors_dtors
6821 if HAVE_INITFINI_ARRAY_SUPPORT.
6822 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
6823 if HAVE_INITFINI_ARRAY_SUPPORT.
6825 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
6827 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
6828 default search path for VxWorks < 7.
6830 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
6832 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
6833 (get_format_string): Refer to c_getstr.
6835 2018-08-21 Tom de Vries <tdevries@suse.de>
6837 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
6838 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
6839 (debuginfo_early_stop): Declare.
6840 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
6841 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
6842 (debuginfo_early_stop): New function.
6843 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
6844 and debuginfo_early_stop.
6845 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
6846 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
6847 (general_init): Call debuginfo_early_init.
6848 (finalize): Call debuginfo_fini.
6849 (do_compile): Call debuginfo_init.
6850 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
6852 (@item -fdump-debug, @item -fdump-earlydebug): Add.
6854 2018-08-21 Tom de Vries <tdevries@suse.de>
6856 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
6857 flag_dump_noaddr and flag_dump_unnumbered.
6859 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
6861 * wide-int-range.cc (wide_int_range_abs): New.
6862 (wide_int_range_order_set): Rename from wide_int_range_min_max.
6863 * wide-int-range.h (wide_int_range_abs): New.
6864 (wide_int_range_min_max): New.
6865 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
6866 case to call wide_int_range_abs.
6867 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
6868 (extract_range_from_abs_expr): Delete.
6870 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
6873 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
6874 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
6877 2018-08-20 Nathan Sidwell <nathan@acm.org>
6878 Jeff Law <law@redhat.com>
6880 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
6881 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
6883 2018-08-20 David Malcolm <dmalcolm@redhat.com>
6886 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
6887 (decl_attributes): Likewise.
6888 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
6890 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
6891 * diagnostic-core.h (class auto_diagnostic_group): New class.
6892 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
6893 (diagnostic_report_diagnostic): Handle the first diagnostics within
6895 (emit_diagnostic): Add auto_diagnostic_group instance.
6897 (inform_n): Likewise.
6898 (warning): Likewise.
6899 (warning_at): Likewise.
6900 (warning_n): Likewise.
6901 (pedwarn): Likewise.
6902 (permerror): Likewise.
6904 (error_n): Likewise.
6905 (error_at): Likewise.
6907 (fatal_error): Likewise.
6908 (internal_error): Likewise.
6909 (internal_error_no_backtrace): Likewise.
6910 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
6911 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
6912 * diagnostic.h (struct diagnostic_context): Add fields
6913 "diagnostic_group_nesting_depth",
6914 "diagnostic_group_emission_count", "begin_group_cb",
6916 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
6917 Add auto_diagnostic_group instance(s).
6918 (find_explicit_erroneous_behavior): Likewise.
6919 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
6920 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
6921 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
6922 (gimplify_va_arg_expr): Likewise.
6923 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
6924 (HSA_SORRY_AT): Likewise.
6925 * ipa-devirt.c (compare_virtual_tables): Likewise.
6926 (warn_odr): Likewise.
6927 * multiple_target.c (expand_target_clones): Likewise.
6928 * opts-common.c (cmdline_handle_error): Likewise.
6929 * reginfo.c (globalize_reg): Likewise.
6930 * substring-locations.c (format_warning_n_va): Likewise.
6931 * tree-inline.c (expand_call_inline): Likewise.
6932 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
6933 * tree-ssa-loop-niter.c
6934 (do_warn_aggressive_loop_optimizations): Likewise.
6935 * tree-ssa-uninit.c (warn_uninit): Likewise.
6936 * tree.c (warn_deprecated_use): Likewise.
6938 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
6941 * config/i386/i386.md (eh_return): Always update EH return
6942 address in word_mode.
6944 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
6946 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
6947 TARGET_SPLIT_COMPLEX_ARG is defined.
6949 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
6951 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
6953 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
6956 * expr.c (expand_assignment): Assert that bitpos is positive.
6957 (store_field): Likewise
6958 (expand_expr_real_1): Make sure that bitpos is positive.
6959 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
6962 2018-08-20 Nathan Sidwell <nathan@acm.org>
6964 * Makefile.in (CPP_ID_DATA_H): Delete.
6965 (CPP_INTERNAL_H): Don't add it.
6966 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
6967 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
6969 2018-08-20 Richard Biener <rguenther@suse.de>
6971 PR tree-optimization/78655
6972 * tree-vrp.c (extract_range_from_binary_expr_1): Make
6973 pointer + offset nonnull if either operand is nonnull work.
6975 2018-08-20 Tom de Vries <tdevries@suse.de>
6977 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
6978 unless the referenced die describes the added property using
6979 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
6980 Otherwise, add a DW_AT_location to the referenced die.
6982 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
6985 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
6986 register_operand when calling ix86_set_reg_reg_cost.
6987 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
6988 Set *total to 0 for operands that satisfy x86_64_immediate_operand
6989 predicate and to 1 otherwise.
6991 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
6993 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
6994 emit a diagnostic that it is not supported and reset the option.
6995 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
6996 supported and consume the option. (ASM_FINAL_SPEC): New.
6998 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
7000 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
7003 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
7005 C-SKY port: Documentation
7007 * doc/extend.texi (C-SKY Function Attributes): New section.
7008 * doc/invoke.texi (Option Summary): Add C-SKY options.
7009 (C-SKY Options): New section.
7010 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
7012 2018-08-17 Jojo <jijie_rong@c-sky.com>
7013 Huibin Wang <huibin_wang@c-sky.com>
7014 Sandra Loosemore <sandra@codesourcery.com>
7015 Chung-Lin Tang <cltang@codesourcery.com>
7017 C-SKY port: Backend implementation
7019 * config/csky/*: New.
7020 * common/config/csky/*: New.
7022 2018-08-17 Jojo <jijie_rong@c-sky.com>
7023 Huibin Wang <huibin_wang@c-sky.com>
7024 Sandra Loosemore <sandra@codesourcery.com>
7025 Chung-Lin Tang <cltang@codesourcery.com>
7026 Andrew Jenner <andrew@codesourcery.com>
7028 C-SKY port: Configury
7030 * config.gcc (csky-*-*): New.
7031 * configure.ac: Add csky to targets for dwarf2 debug_line support.
7032 * configure: Regenerated.
7034 2018-08-17 David Malcolm <dmalcolm@redhat.com>
7036 * dump-context.h: Include "dumpfile.h".
7037 (dump_context::dump_printf_va): Convert final param from va_list
7038 to va_list *. Convert from ATTRIBUTE_PRINTF to
7039 ATTRIBUTE_GCC_DUMP_PRINTF.
7040 (dump_context::dump_printf_loc_va): Likewise.
7041 * dumpfile.c: Include "stringpool.h".
7042 (make_item_for_dump_printf_va): Delete.
7043 (make_item_for_dump_printf): Delete.
7044 (class dump_pretty_printer): New class.
7045 (dump_pretty_printer::dump_pretty_printer): New ctor.
7046 (dump_pretty_printer::emit_items): New member function.
7047 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
7049 (dump_pretty_printer::emit_item): New member function.
7050 (dump_pretty_printer::stash_item): New member function.
7051 (dump_pretty_printer::format_decoder_cb): New member function.
7052 (dump_pretty_printer::decode_format): New member function.
7053 (dump_context::dump_printf_va): Reimplement in terms of
7054 dump_pretty_printer.
7055 (dump_context::dump_printf_loc_va): Convert final param from va_list
7057 (dump_context::begin_scope): Reimplement call to
7058 make_item_for_dump_printf.
7059 (dump_printf): Update for change to dump_printf_va.
7060 (dump_printf_loc): Likewise.
7061 (selftest::test_capture_of_dump_calls): Convert "stmt" from
7062 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
7063 with %T, %E, and %G.
7064 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
7065 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
7066 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
7067 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
7068 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
7069 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
7070 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
7071 within a dump_printf_loc call to "%wu".
7072 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
7073 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
7074 missing space after "=".
7075 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
7076 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
7077 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
7078 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
7079 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
7080 duplicate "vectorized" from message.
7082 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
7084 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
7085 polyNxK_t element's TYPE_STRING_FLAG.
7087 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
7089 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
7090 (they were unnamed before). Fix comments.
7092 2018-08-17 Nathan Sidwell <nathan@acm.org>
7094 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
7096 2018-08-17 Richard Biener <rguenther@suse.de>
7098 PR tree-optimization/86841
7099 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
7101 2018-08-17 Martin Liska <mliska@suse.cz>
7103 * common.opt: Remove Warn, Init and Report for options with
7104 Ignore/Deprecated flag. Warning is done automatically for
7106 * config/i386/i386.opt: Likewise.
7107 * config/ia64/ia64.opt: Likewise.
7108 * config/rs6000/rs6000.opt: Likewise.
7109 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
7110 Remove usage of flag_check_pointer_bounds.
7111 * lto-wrapper.c (merge_and_complain): Do not handle
7112 OPT_fcheck_pointer_bounds.
7113 (append_compiler_options): Likewise.
7114 * opt-functions.awk: Do not handle Deprecated.
7115 * optc-gen.awk: Check that Var, Report and Init are not
7116 used for an option with Ignore/Deprecated flag.
7117 * opts-common.c (decode_cmdline_option): Do not report
7119 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
7121 * opts.h (struct cl_option): Remove cl_deprecated flag.
7122 (CL_ERR_DEPRECATED): Remove error enum value.
7124 2018-08-17 Richard Biener <rguenther@suse.de>
7127 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
7128 across a va-arg-pack using call adjust its return value accordingly.
7130 2018-08-16 Martin Sebor <msebor@redhat.com>
7132 PR tree-optimization/86853
7133 * gimple-ssa-sprintf.c (struct format_result): Rename member.
7134 (struct fmtresult): Add member and initialize it in ctors.
7135 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
7136 (format_string): Handle %S the same as %ls. Set MAYFAIL.
7137 (format_directive): Set POSUNDER4K when MAYFAIL is set.
7138 (parse_directive): Handle %C same as %c.
7139 (sprintf_dom_walker::compute_format_length): Adjust.
7140 (is_call_safe): Adjust.
7142 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
7144 * builtins.c (c_strlen): Add new parameter eltsize. Use it
7145 for determining how to count the elements.
7146 * builtins.h (c_strlen): Adjust prototype.
7147 * expr.c (string_constant): Add new parameter mem_size.
7148 Set *mem_size appropriately.
7149 * expr.h (string_constant): Adjust protoype.
7150 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
7151 * gimple-fold.h (get_range_strlen): Adjust prototype.
7152 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
7153 (format_string): Call get_string_length with eltsize.
7155 2018-08-16 David Malcolm <dmalcolm@redhat.com>
7157 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
7158 to emit the span, rather than setting it as the prefix.
7160 2018-08-16 David Malcolm <dmalcolm@redhat.com>
7162 * diagnostic-show-locus.c (layout::start_annotation_line): Add
7163 "margin_char" parameter, defaulting to space. Use it in place
7164 of pp_space for the initial part of the margin.
7165 (layout::print_leading_fixits): Use '+' when filling the margin
7166 of line-insertion fix-it hints.
7168 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
7170 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
7173 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
7175 * config/rs6000/altivec.md: Don't set length attribute to the default
7177 * config/rs6000/darwin.md: Ditto.
7178 * config/rs6000/dfp.md: Ditto.
7179 * config/rs6000/htm.md: Ditto.
7180 * config/rs6000/rs6000.md: Ditto.
7181 * config/rs6000/sync.md: Ditto.
7182 * config/rs6000/vsx.md: Ditto.
7184 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
7186 * config/rs6000/altivec.md: Don't set length attribute to the default
7187 value, for branch instructions.
7188 * config/rs6000/darwin.md: Ditto.
7189 * config/rs6000/rs6000.md: Ditto.
7191 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
7193 * config/rs6000/rs6000.md (length): Always define as const_int 4.
7194 (unnamed conditional branch define_insn): Set length to 4 or 8
7195 depending on offset.
7196 (<bd>_<mode>): Similar, for alternative 0.
7197 (<bd>tf_<mode>): Ditto.
7199 2018-08-16 Tamar Christina <tamar.christina@arm.com>
7201 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
7203 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
7205 * doc/rtl.texi: Replace old RTX class names with new names.
7208 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
7210 * expmed.h (canonicalize_comparison): New declaration.
7211 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
7212 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
7213 * optabs.c (prepare_cmp_insn): Likewise.
7214 * rtl.h (unsigned_condition_p): New function which checks if a
7215 comparison operator is unsigned.
7217 2018-08-16 Nathan Sidwell <nathan@acm.org>
7219 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
7220 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
7222 2018-08-16 Tamar Christina <tamar.christina@arm.com>
7225 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
7226 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
7227 (mov<mov>): ..this and enable unconditionally.
7229 2018-08-16 Tamar Christina <tamar.christina@arm.com>
7231 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
7233 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
7235 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
7236 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
7237 "Common" with "Target".
7239 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
7241 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
7242 * doc/invoke.texi (mmitigate-rop): Remove.
7243 * config/i386/i386.c: Do not include "regrename.h".
7244 (ix86_rop_should_change_byte_p, reg_encoded_number)
7245 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
7247 (ix86_reorg): Remove call to ix86_mitigate_rop.
7248 * config/i386/i386.md (attr "modrm_class"): Remove.
7249 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
7250 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
7251 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
7253 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
7255 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
7256 allow folding of mergeh() and mergel() for the float and double types.
7257 (fold_mergehl_helper): Rework to handle building a permute tree
7260 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
7262 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
7265 2018-08-15 David Malcolm <dmalcolm@redhat.com>
7267 * common.opt (fdiagnostics-show-labels): New option.
7268 * diagnostic-show-locus.c (class layout_range): Add field
7270 (class layout): Add field "m_show_labels_p".
7271 (layout_range::layout_range): Add param "label" and use it to
7273 (make_range): Pass in NULL for new "label" param of layout_range's
7275 (layout::layout): Initialize m_show_labels_p.
7276 (layout::maybe_add_location_range): Pass in loc_range->m_label
7277 when constructing layout_range instances.
7278 (struct line_label): New struct.
7279 (layout::print_any_labels): New member function.
7280 (layout::print_line): Call it if label-printing is enabled.
7281 (selftest::test_one_liner_labels): New test.
7282 (selftest::test_diagnostic_show_locus_one_liner): Call it.
7283 * diagnostic.c (diagnostic_initialize): Initialize
7284 context->show_labels_p.
7285 * diagnostic.h (struct diagnostic_context): Add field
7287 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
7288 -fno-diagnostics-show-labels.
7289 * dwarf2out.c (gen_producer_string): Add
7290 OPT_fdiagnostics_show_labels to the ignored options.
7291 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
7293 (gcc_rich_location::maybe_add_expr): Likewise.
7294 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
7295 label" param, defaulting to NULL.
7296 (gcc_rich_location::add_expr): Add "label" param.
7297 (gcc_rich_location::maybe_add_expr): Likewise.
7298 (class text_range_label): New class.
7299 (class range_label_for_type_mismatch): New class.
7300 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
7301 of format_warning_va.
7302 (fmtwarn_n): Likewise for new params of format_warning_n_va.
7303 * lto-wrapper.c (merge_and_complain): Add
7304 OPT_fdiagnostics_show_labels to the "pick one setting" options.
7305 (append_compiler_options): Likewise to the dropped options.
7306 (append_diag_options): Likewise to the passed-on options.
7307 * opts.c (common_handle_option): Handle the new option.
7308 * selftest-diagnostic.c
7309 (test_diagnostic_context::test_diagnostic_context): Enable
7311 * substring-locations.c: Include "gcc-rich-location.h".
7312 (format_warning_n_va): Add "fmt_label" and "param_label" params
7313 and use them as appropriate.
7314 (format_warning_va): Add "fmt_label" and "param_label" params,
7315 passing them on to format_warning_n_va.
7316 (format_warning_at_substring): Likewise.
7317 (format_warning_at_substring_n): Likewise.
7318 * substring-locations.h (format_warning_va): Add "fmt_label" and
7319 "param_label" params.
7320 (format_warning_n_va): Likewise.
7321 (format_warning_at_substring): Likewise.
7322 (format_warning_at_substring_n): Likewise.
7323 * toplev.c (general_init): Initialize global_dc->show_labels_p.
7325 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
7328 * builtins.c (expand_builtin_memcmp): Do not expand the call
7329 when overflow is detected.
7331 2018-08-15 Martin Sebor <msebor@redhat.com>
7333 PR tree-optimization/71625
7334 * config/aarch64/aarch64-builtins.c
7335 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
7337 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
7339 * config/s390/s390.c (s390_reorg): Remove loop.
7341 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
7344 (darwin_function_switched_text_sections): Delete.
7345 * gcc/config/darwin.h
7346 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
7348 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
7351 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
7352 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
7353 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
7355 2018-08-15 Martin Liska <mliska@suse.cz>
7357 PR tree-optimization/86925
7358 * predict.c (expr_expected_value_1): When taking
7359 later predictor, assign also probability.
7360 Use fold_build2_initializer_loc in order to fold
7361 the expression in -frounding-math.
7363 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
7365 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
7367 (expand_vec_perm_1): Try the new method.
7369 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
7372 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
7373 Check whether lra_live_max_point is 0 before dividing.
7375 2018-08-14 Martin Sebor <msebor@redhat.com>
7377 PR tree-optimization/86650
7378 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
7379 (vrp_prop::check_mem_ref): Same.
7381 2018-08-13 Liu Hao <lh_mouse@126.com>
7383 * pretty-print.c (eat_esc_sequence): Swap the foreground and
7384 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
7385 and clear it thereafter, as it only works for DBCS.
7387 2018-08-13 Liu Hao <lh_mouse@126.com>
7389 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
7390 handle returned by _get_osf_handle().
7392 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
7394 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
7395 for folding vec_perm.
7397 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
7399 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
7400 Add support for gimple-folding of vec_pack() and vec_unpack()
7403 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
7405 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
7406 vec_xst variants to the list.
7407 (rs6000_gimple_fold_builtin): Add support for folding unaligned
7408 vector loads and stores.
7410 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
7412 * config.gcc (rs6000-ibm-aix4.x): Delete.
7413 (rs6000-ibm-aix5.1): Delete.
7414 (rs6000-ibm-aix5.2): Delete.
7415 (rs6000-ibm-aix5.3): Delete.
7416 * config/rs6000/aix43.h: Delete.
7417 * config/rs6000/aix51.h: Delete.
7418 * config/rs6000/aix52.h: Delete.
7419 * config/rs6000/t-aix43: Delete.
7421 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
7423 * config/s390/s390.c (s390_decompose_constant_pool_ref):
7425 (s390_decompose_address): Factor out constant pool ref
7428 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
7430 * config/nds32/nds32-predicates.c
7431 (nds32_can_use_bclr_p): Change return type as bool.
7432 (nds32_can_use_bset_p): Ditto.
7433 (nds32_can_use_btgl_p): Ditto.
7434 (nds32_can_use_bitci_p): Ditto.
7435 * config/nds32/nds32-protos.h
7436 (nds32_can_use_bclr_p): Change declaration.
7437 (nds32_can_use_bset_p): Ditto.
7438 (nds32_can_use_btgl_p): Ditto.
7439 (nds32_can_use_bitci_p): Ditto.
7441 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
7443 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
7444 Support -msched-prolog-epilog option.
7445 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
7447 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
7449 * common/config/nds32/nds32-common.c
7450 (nds32_option_optimization_table): Enalbe -malways-align.
7452 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
7454 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
7456 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
7457 OPT_misr_secure_ case.
7458 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
7459 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
7460 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
7462 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
7463 (nds32_isr_info): New field security_level.
7464 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
7465 * config/nds32/nds32.md (return_internal): Consider critical attribute.
7466 * config/nds32/nds32.opt (misr-secure): New option.
7467 * config/nds32/nds32_init.inc: New file.
7468 * config/nds32/nds32_isr.h: New file.
7470 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
7472 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
7473 Update comment for atomic instructions.
7474 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
7475 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
7477 (atomic_loaddi): Revise fence expansion to only emit fence prior to
7478 load for __ATOMIC_SEQ_CST model.
7479 (atomic_loaddi_1): Remove float register target.
7480 (atomic_storedi): Handle CONST_INT values.
7481 (atomic_storedi_1): Remove float register source. Add special case
7483 (memory_barrier): New expander and insn.
7485 2018-08-11 Jakub Jelinek <jakub@redhat.com>
7487 PR tree-optimization/86835
7488 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
7489 new_stmt after def_gsi, make sure to insert new_square_stmt after
7490 that stmt, not 2 stmts before it.
7492 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
7495 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
7498 2018-08-10 Martin Liska <mliska@suse.cz>
7501 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
7503 * builtins.c (expand_builtin_expect_with_probability):
7505 (expand_builtin_expect_with_probability): New function.
7506 (build_builtin_expect_predicate): Add new argumnet probability
7507 for BUILT_IN_EXPECT_WITH_PROBABILITY.
7508 (fold_builtin_expect):
7511 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
7512 * builtins.h (fold_builtin_expect): Set new argument.
7513 * doc/extend.texi: Document __builtin_expect_with_probability.
7514 * doc/invoke.texi: Likewise.
7515 * gimple-fold.c (gimple_fold_call): Pass new argument.
7516 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
7517 also BUILT_IN_EXPECT_WITH_PROBABILITY.
7518 * predict.c (get_predictor_value): New function.
7519 (expr_expected_value): Add new argument probability. Assume
7520 that predictor and probability are always non-null.
7521 (expr_expected_value_1): Likewise. For __builtin_expect and
7522 __builtin_expect_with_probability set probability. Handle
7523 combination in binary expressions.
7524 (tree_predict_by_opcode): Simplify code by simply calling
7525 get_predictor_value.
7526 (pass_strip_predict_hints::execute): Add handling of
7527 BUILT_IN_EXPECT_WITH_PROBABILITY.
7528 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
7530 * tree.h (DECL_BUILT_IN_P): New function.
7532 2018-08-10 Martin Liska <mliska@suse.cz>
7534 PR tree-optimization/85799
7535 * passes.def: Add argument for pass_strip_predict_hints.
7536 * predict.c (class pass_strip_predict_hints): Add new argument
7538 (strip_predictor_early): New function.
7539 (pass_strip_predict_hints::execute): Call the function to
7541 (strip_predict_hints): New function.
7542 * predict.def: Fix comment.
7544 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
7546 * Makefile.in: Clarify which tm.texi to copy over to assert the
7547 right to grant a GFDL license for all.
7549 2018-08-09 Jeff Law <law@redhat.com>
7551 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
7554 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
7556 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
7559 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
7561 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
7562 reductions for variable-length vectors.
7564 2018-08-09 David Malcolm <dmalcolm@redhat.com>
7567 * common.opt (fdiagnostics-show-line-numbers): New option.
7568 * diagnostic-show-locus.c (class layout): Add fields
7569 "m_show_line_numbers_p" and "m_linenum_width";
7570 (num_digits): New function.
7571 (test_num_digits): New function.
7572 (layout::layout): Initialize new fields. Update m_x_offset
7573 logic to handle any left margin.
7574 (layout::print_source_line): Print line number when requested.
7575 (layout::start_annotation_line): New member function.
7576 (layout::print_annotation_line): Call it.
7577 (layout::print_leading_fixits): Likewise.
7578 (layout::print_trailing_fixits): Likewise. Update calls to
7579 move_to_column for new parameter.
7580 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
7581 it to potentially call start_annotation_line.
7582 (layout::show_ruler): Call start_annotation_line.
7583 (selftest::test_line_numbers_multiline_range): New selftest.
7584 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
7585 and selftest::test_line_numbers_multiline_range.
7586 * diagnostic.c (diagnostic_initialize): Initialize
7587 show_line_numbers_p.
7588 * diagnostic.h (struct diagnostic_context): Add field
7589 "show_line_numbers_p".
7590 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
7591 -fno-diagnostics-show-line-numbers.
7592 * dwarf2out.c (gen_producer_string): Add
7593 OPT_fdiagnostics_show_line_numbers to the ignored options.
7594 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
7595 one setting" options.
7596 (append_compiler_options): Likewise to the dropped options.
7597 (append_diag_options): Likewise to the passed-on options.
7598 * opts.c (common_handle_option): Handle the new option.
7599 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
7601 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
7603 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
7604 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
7605 third argument of type "const signed char" to descriptions of
7606 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
7607 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
7608 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
7609 __builtin_bcdsub_ov functions.
7611 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
7613 PR tree-optimization/86858
7614 * tree-vect-loop.c (vect_is_simple_reduction): Restore
7615 flow_bb_inside_loop_p calls.
7617 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
7619 PR tree-optimization/86871
7620 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
7621 instead of gimple_assign_lhs.
7623 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
7626 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
7627 register constraint to operand 0.
7628 (add<mode>3_carryinC): Likewise.
7629 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
7631 2018-08-09 Martin Liska <mliska@suse.cz>
7634 * common.opt: Remove extra line.
7636 2018-08-09 Martin Liska <mliska@suse.cz>
7638 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
7639 at the end of a line, make first letter capital and end up
7640 a sentence with a dot.
7641 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
7642 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
7643 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
7644 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
7645 (PARAM_MAX_ISL_OPERATIONS): Likewise.
7646 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
7647 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
7648 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
7649 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
7650 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
7651 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
7652 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
7653 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
7654 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
7655 (PARAM_TREE_REASSOC_WIDTH): Likewise.
7656 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
7657 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
7658 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
7660 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
7663 * config/s390/s390.c (s390_option_override_internal): Reduce the
7664 stack-clash-protection-probe-interval param if it would be too big
7667 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
7670 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
7671 * config/m68k/m68k.c (handle_move_double): Don't call
7672 m68k_final_prescan_insn.
7673 (m68k_adjust_decorated_operand): Renamed from
7674 m68k_final_prescan_insn, remove first and third operand and
7676 (print_operand): Call it.
7677 (print_operand_address): Call it.
7679 2018-08-08 Nathan Sidwell <nathan@acm.org>
7681 * diagnostic.c (diagnostic_report_current_module): Use
7682 linemap_included_from & linemap_included_from_linemap.
7684 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
7686 * config/aarch64/aarch64-cores.def: Add phecda core.
7687 * config/aarch64/aarch64-tune.md: Regenerate.
7688 * doc/invoke.texi: Add phecda core.
7690 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
7693 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
7695 * config/s390/s390.md ("movti"): Add more alternatives for
7696 constant to GPR copies.
7698 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
7700 * config/s390/s390.c: Fix whitespace damage throughout the file.
7701 * config/s390/s390.h: Likewise.
7702 * config/s390/tpf.h: Likewise.
7704 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
7706 * config/s390/s390.c (s390_loadrelative_operand_p):
7707 Remove TARGET_CPU_ZARCH usages.
7708 (s390_rtx_costs): Likewise.
7709 (s390_legitimate_constant_p): Likewise.
7710 (s390_cannot_force_const_mem): Likewise.
7711 (legitimate_reload_constant_p): Likewise.
7712 (s390_preferred_reload_class): Likewise.
7713 (legitimize_pic_address): Likewise.
7714 (legitimize_tls_address): Likewise.
7715 (s390_split_branches): Removed.
7716 (s390_add_execute): Removed.
7717 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
7718 (s390_mainpool_start): Likewise.
7719 (s390_mainpool_finish): Likewise.
7720 (s390_mainpool_cancel): Removed.
7721 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
7722 (s390_chunkify_cancel): Likewise.
7723 (s390_return_addr_rtx): Likewise.
7724 (s390_register_info): Remove split_branches_pending_p uages.
7725 (s390_optimize_register_info): Likewise.
7726 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
7727 split_branches_pending_p usages.
7728 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
7729 (s390_load_got): Likewise.
7730 (s390_expand_split_stack_prologue): Likewise.
7731 (output_asm_nops): Likewise.
7732 (s390_function_profiler): Likewise.
7733 (s390_emit_call): Likewise.
7734 (s390_conditional_register_usage): Likewise.
7735 (s390_optimize_prologue): Likewise.
7736 (s390_reorg): Remove TARGET_CPU_ZARCH and
7737 split_branches_pending_p usages.
7738 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
7740 (s390_output_indirect_thunk_function): Likewise.
7741 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
7742 (TARGET_CPU_ZARCH_P): Removed.
7743 (struct machine_function): Remove split_branches_pending_p.
7744 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
7746 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
7748 * common/config/s390/s390-common.c (processor_flags_table):
7750 * config.gcc: Remove with_arch/with_tune support.
7751 * config/s390/2064.md: Remove cpu attribute comparisons.
7752 * config/s390/driver-native.c (s390_host_detect_local_cpu):
7754 * config/s390/linux.h (ASM_SPEC):
7755 Remove -march support.
7756 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
7757 Use a table to get an arch level.
7758 * config/s390/s390-opts.h (enum processor_type):
7760 * config/s390/s390.c
7761 (processor_table): Remove entries, add arch_level values.
7762 (s390_issue_rate): Remove cases.
7763 (s390_option_override): Adjust
7764 s390_option_override_internal() call.
7765 (s390_option_override_internal): Remove deprecation warning.
7766 (s390_valid_target_attribute_tree): Adjust
7767 s390_option_override_internal() call.
7768 * config/s390/s390.h (struct s390_processor):
7769 Share with s390-c.c, add arch_level field.
7770 * config/s390/s390.md:
7771 Remove occurrences in cpu attribute.
7772 * config/s390/s390.opt: Remove -march/-mtune support.
7773 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
7774 * doc/invoke.texi: Remove deprecation warning.
7776 2018-08-08 Luis Machado <luis.machado@linaro.org>
7778 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
7780 (qdf24xx_tunings): Set vector cost structure to
7781 qdf24xx_vector_cost.
7783 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
7784 <register_sextend>: Set to 3.
7786 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
7789 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
7790 * config/aarch64/aarch64-simd.md
7791 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
7792 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
7793 (aarch64_frecpx<mode>): ...this new pattern.
7794 * config/aarch64/aarch64-simd-builtins.def: Remove comment
7795 about aarch64_frecp<FRECP:frecp_suffix><mode>.
7797 2018-08-07 Martin Liska <mliska@suse.cz>
7800 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
7801 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
7802 * predict.def (PRED_MALLOC_NONNULL): New predictor.
7803 * doc/extend.texi: Document that malloc attribute adds
7806 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
7809 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7810 Define to speculation_safe_value_not_needed.
7812 2018-08-06 Jeff Law <law@redhat.com>
7814 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
7815 the vr_values instance to cprop_into_stmt.
7816 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
7817 (cprop_operand): Also query EVRP to determine if OP is a constant.
7819 2018-08-06 Nathan Sidwell <nathan@acm.org>
7821 * diagnostic.c (diagnostic_report_current_module): Reroll
7822 included-at loop. Translate text.
7824 2018-08-06 David Malcolm <dmalcolm@redhat.com>
7826 * function-tests.c (selftest::test_expansion_to_rtl): Call
7827 free_after_compilation.
7829 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7831 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
7833 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
7835 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
7836 loops with memory block operations from getting unrolled.
7838 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
7841 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7842 Define to speculation_safe_value_not_needed.
7844 2018-08-06 Jeff Law <law@redhat.com>
7846 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
7849 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
7852 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
7853 with all enabled __intN types.
7855 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
7857 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7859 * alias.c (record_set): Check for clobber high.
7860 * cfgexpand.c (expand_gimple_stmt): Likewise.
7861 * combine-stack-adj.c (single_set_for_csa): Likewise.
7862 * combine.c (find_single_use_1): Likewise.
7863 (set_nonzero_bits_and_sign_copies): Likewise.
7864 (get_combine_src_dest): Likewise.
7865 (is_parallel_of_n_reg_sets): Likewise.
7866 (try_combine): Likewise.
7867 (record_dead_and_set_regs_1): Likewise.
7868 (reg_dead_at_p_1): Likewise.
7869 (reg_dead_at_p): Likewise.
7870 * dce.c (deletable_insn_p): Likewise.
7871 (mark_nonreg_stores_1): Likewise.
7872 (mark_nonreg_stores_2): Likewise.
7873 * df-scan.c (df_find_hard_reg_defs): Likewise.
7874 (df_uses_record): Likewise.
7875 (df_get_call_refs): Likewise.
7876 * dwarf2out.c (mem_loc_descriptor): Likewise.
7877 * haifa-sched.c (haifa_classify_rtx): Likewise.
7878 * ira-build.c (create_insn_allocnos): Likewise.
7879 * ira-costs.c (scan_one_insn): Likewise.
7880 * ira.c (equiv_init_movable_p): Likewise.
7881 (rtx_moveable_p): Likewise.
7882 (interesting_dest_for_shprep): Likewise.
7883 * jump.c (mark_jump_label_1): Likewise.
7884 * postreload-gcse.c (record_opr_changes): Likewise.
7885 * postreload.c (reload_cse_simplify): Likewise.
7886 (struct reg_use): Add source expr.
7887 (reload_combine): Check for clobber high.
7888 (reload_combine_note_use): Likewise.
7889 (reload_cse_move2add): Likewise.
7890 (move2add_note_store): Likewise.
7891 * print-rtl.c (print_pattern): Likewise.
7892 * recog.c (decode_asm_operands): Likewise.
7893 (store_data_bypass_p): Likewise.
7894 (if_test_bypass_p): Likewise.
7895 * regcprop.c (kill_clobbered_value): Likewise.
7896 (kill_set_value): Likewise.
7897 * reginfo.c (reg_scan_mark_refs): Likewise.
7898 * reload1.c (maybe_fix_stack_asms): Likewise.
7899 (eliminate_regs_1): Likewise.
7900 (elimination_effects): Likewise.
7901 (mark_not_eliminable): Likewise.
7902 (scan_paradoxical_subregs): Likewise.
7903 (forget_old_reloads_1): Likewise.
7904 * reorg.c (find_end_label): Likewise.
7905 (try_merge_delay_insns): Likewise.
7906 (redundant_insn): Likewise.
7907 (own_thread_p): Likewise.
7908 (fill_simple_delay_slots): Likewise.
7909 (fill_slots_from_thread): Likewise.
7910 (dbr_schedule): Likewise.
7911 * resource.c (update_live_status): Likewise.
7912 (mark_referenced_resources): Likewise.
7913 (mark_set_resources): Likewise.
7914 * rtl.c (copy_rtx): Likewise.
7915 * rtlanal.c (reg_referenced_p): Likewise.
7916 (single_set_2): Likewise.
7917 (noop_move_p): Likewise.
7918 (note_stores): Likewise.
7919 * sched-deps.c (sched_analyze_reg): Likewise.
7920 (sched_analyze_insn): Likewise.
7922 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7924 * cse.c (invalidate_reg): New function extracted from...
7925 (invalidate): ...here.
7926 (canonicalize_insn): Check for clobber high.
7927 (invalidate_from_clobbers): invalidate clobber highs.
7928 (invalidate_from_sets_and_clobbers): Likewise.
7929 (count_reg_usage): Check for clobber high.
7930 (insn_live_p): Likewise.
7931 * cselib.c (cselib_expand_value_rtx_1):Likewise.
7932 (cselib_invalidate_regno): Check for clobber in setter.
7933 (cselib_invalidate_rtx): Pass through setter.
7934 (cselib_invalidate_rtx_note_stores):
7935 (cselib_process_insn): Check for clobber high.
7936 * cselib.h (cselib_invalidate_rtx): Add operand.
7938 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7940 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
7941 (mark_not_eliminable): Likewise.
7942 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
7943 * lra-lives.c (process_bb_lives): Check for clobber high.
7944 * lra.c (new_insn_reg): Remember clobber highs.
7945 (collect_non_operand_hard_regs): Check for clobber high.
7946 (lra_set_insn_recog_data): Likewise.
7947 (add_regs_to_insn_regno_info): Likewise.
7948 (lra_update_insn_regno_info): Likewise.
7950 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7952 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
7953 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
7955 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7957 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
7958 (copy_insn_1): Likewise.
7959 (gen_hard_reg_clobber_high): New gen function.
7960 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
7961 * genemit.c (gen_exp): Likewise.
7962 (gen_emit_seq): Pass through info.
7963 (gen_insn): Check for CLOBBER_HIGH.
7964 (gen_expand): Pass through info.
7965 (gen_split): Likewise.
7966 (output_add_clobbers): Likewise.
7967 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
7968 (remove_clobbers): Likewise.
7969 * rtl.h (gen_hard_reg_clobber_high): New declaration.
7971 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7973 * doc/rtl.texi (clobber_high): Add.
7974 (parallel): Add in clobber high
7975 * rtl.c (rtl_check_failed_code3): Add function.
7976 * rtl.def (CLOBBER_HIGH): Add expression.
7977 * rtl.h (RTL_CHECKC3): Add macro.
7978 (rtl_check_failed_code3): Add declaration.
7979 (XC3EXP): Add macro.
7981 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
7984 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
7985 cfun->machine->max_used_stack_alignment if needed.
7987 2018-08-04 Martin Sebor <msebor@redhat.com>
7989 PR tree-optimization/86571
7990 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
7993 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
7996 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7999 2018-08-03 Jeff Law <law@redhat.com>
8002 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
8003 Define to speculation_safe_value_not_needed.
8005 2018-08-03 David Malcolm <dmalcolm@redhat.com>
8007 * doc/gcov.texi (-x): Remove duplicate "to".
8008 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
8009 (-Wif-not-aligned): Remove duplicate "is".
8010 (-flto): Remove duplicate "the".
8011 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
8012 duplicate "v5.00.b".
8013 (MSP430 Options): Remove duplicate "and" from the description
8014 of "-mgprel-sec=regexp".
8015 (x86 Options): Remove duplicate copies of "vmldLog102" and
8016 vmlsLog104 from description of "-mveclibabi=type".
8018 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
8020 * internal-fn.h (first_commutative_argument): Declare.
8021 * internal-fn.c (first_commutative_argument): New function.
8022 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
8023 restrictions for pattern statements. Use first_commutative_argument
8024 to look for commutative operands in calls to internal functions.
8026 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
8028 * Makefile.in (wide-int-range.o): New.
8029 * tree-vrp.c: Move all the wide_int_* functions to...
8030 * wide-int-range.cc: ...here.
8031 * tree-vrp.h: Move all the wide_int_* prototypes to...
8032 * wide-int-range.h: ...here.
8034 2018-08-03 Tom de Vries <tdevries@suse.de>
8036 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
8038 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
8039 * except.c (output_function_exception_table): Do early exit if
8040 targetm_common.except_unwind_info (&global_options) == UI_NONE.
8042 2018-08-03 Martin Liska <mliska@suse.cz>
8044 * predict.c (dump_prediction): Change to 2 digits
8047 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
8049 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
8051 (falkor_am_1_gtov_gtov): ... a new insn reservation.
8053 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
8055 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
8056 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
8057 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
8058 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
8059 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
8061 2018-08-02 David Malcolm <dmalcolm@redhat.com>
8063 * diagnostic-show-locus.c (diagnostic_show_locus): Use
8064 pp_take_prefix when saving the existing prefix.
8065 * diagnostic.c (diagnostic_append_note): Likewise.
8066 * langhooks.c (lhd_print_error_function): Likewise.
8067 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
8068 param's type. Free the existing prefix.
8069 (pp_take_prefix): New function.
8070 (pretty_printer::pretty_printer): Drop the prefix parameter.
8071 Rename the length parameter to match the comment.
8072 (pretty_printer::~pretty_printer): Free the prefix.
8073 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
8075 (struct pretty_printer): Drop the "const" from "prefix" field's
8076 type and clarify memory management.
8077 (pp_set_prefix): Drop the "const" from the 2nd param.
8078 (pp_take_prefix): New decl.
8080 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
8082 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
8083 for word_mode_ok here instead of passing as argument.
8084 (expand_block_compare): Change select_block_compare_mode() call.
8085 (expand_strncmp_gpr_sequence): New function.
8086 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
8088 2018-08-02 Jeff Law <law@redhat.com>
8091 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
8092 Define to speculation_safe_value_not_needed.
8095 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
8096 Define to speculation_safe_value_not_needed.
8098 2018-08-02 Tom de Vries <tdevries@suse.de>
8101 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
8102 function. Return UI_TARGET unconditionally.
8103 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
8104 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
8106 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
8108 * genemit.c (print_overload_test): Fix typo.
8110 2018-08-02 Richard Biener <rguenther@suse.de>
8112 PR tree-optimization/86816
8113 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
8114 which checks for value availability before querying it.
8115 (gvn_uses_equal): Use it.
8116 (same_succ_hash): Likewise.
8117 (gimple_equal_p): Likewise.
8119 2018-08-02 Nick Clifton <nickc@redhat.com>
8122 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
8123 Define to speculation_safe_value_not_needed.
8126 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
8127 Define to speculation_safe_value_not_needed.
8130 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
8131 Define to speculation_safe_value_not_needed.
8134 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
8135 Define to speculation_safe_value_not_needed.
8138 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
8139 Define to speculation_safe_value_not_needed.
8142 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
8143 Define to speculation_safe_value_not_needed.
8146 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
8147 Define to speculation_safe_value_not_needed.
8150 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
8151 Define to speculation_safe_value_not_needed.
8154 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
8155 speculation_safe_value_not_needed.
8158 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
8159 to speculation_safe_value_not_needed.
8161 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
8163 * doc/md.texi: Expand the documentation of instruction names
8164 to mention port-local uses. Document '@' in pattern names.
8165 * read-md.h (overloaded_instance, overloaded_name): New structs.
8167 (md_reader::handle_overloaded_name): New member function.
8168 (md_reader::get_overloads): Likewise.
8169 (md_reader::m_first_overload): New member variable.
8170 (md_reader::m_next_overload_ptr): Likewise.
8171 (md_reader::m_overloads_htab): Likewise.
8172 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
8173 m_next_overload_ptr and m_overloads_htab.
8174 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
8175 (get_mode_token, get_code_token, get_int_token): New functions.
8176 (map_attr_string): Add an optional argument that passes back
8177 the associated iterator.
8178 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
8179 (md_reader::handle_overloaded_name, add_overload_instance): New
8181 (apply_iterators): Handle '@' names. Report an error if '@'
8182 is used without iterators.
8183 (initialize_iterators): Initialize the new iterator_group fields.
8184 * genopinit.c (handle_overloaded_code_for)
8185 (handle_overloaded_gen): New functions.
8186 (main): Use them to print declarations of maybe_code_for_* and
8187 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
8188 * genemit.c (print_overload_arguments, print_overload_test)
8189 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
8190 (main): Use it to print definitions of maybe_code_for_* and
8191 maybe_gen_* functions.
8192 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
8193 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
8194 instead of explicit mode checks.
8195 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
8196 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
8197 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
8198 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
8199 (aarch64_expand_compare_and_swap): Likewise
8200 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
8201 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
8202 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
8203 (aarch64_constant_pool_reload_icode): Delete.
8204 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
8205 instead of aarch64_constant_pool_reload_icode. Use
8206 code_for_aarch64_reload_mov instead of explicit mode checks.
8207 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
8208 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
8209 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
8210 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
8211 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
8212 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
8213 (aarch64_atomic_load_op_code): Delete.
8214 (aarch64_emit_atomic_load_op): Likewise.
8215 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
8216 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
8217 instead of aarch64_emit_atomic_load_op.
8218 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
8219 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
8220 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
8221 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
8222 character before the pattern name.
8223 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
8224 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
8225 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
8226 (aarch64_frecps<mode>): Likewise.
8227 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
8228 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
8229 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
8230 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
8231 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
8233 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
8235 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
8236 Allow HFmode constants if TARGET_FP_F16INST.
8238 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
8241 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
8242 No longer check last store for clobber of address register.
8244 2018-08-02 Martin Liska <mliska@suse.cz>
8246 PR gcov-profile/86817
8247 * gcov.c (process_all_functions): New function.
8249 (process_file): Move functions processing to
8250 process_all_functions.
8252 2018-08-02 David Malcolm <dmalcolm@redhat.com>
8254 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
8255 "const" to the "gimple *" and "rtx_insn *" parameters.
8256 * dumpfile.h (dump_user_location_t::dump_user_location_t):
8258 (dump_location_t::dump_location_t): Likewise.
8260 2018-08-01 Martin Sebor <msebor@redhat.com>
8262 PR tree-optimization/86650
8263 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
8264 rather than a "gcall *". Directly pass the data of interest
8265 to percent_K_format, rather than building a temporary CALL_EXPR
8267 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
8268 (gimple_fold_builtin_strncat): Adjust.
8269 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
8270 gcall* argument with gimple*.
8271 * gimple-ssa-warn-restrict.c (check_call): Same.
8272 (wrestrict_dom_walker::before_dom_children): Same.
8273 (builtin_access::builtin_access): Same.
8274 (check_bounds_or_overlap): Same
8275 (maybe_diag_overlap): Same.
8276 (maybe_diag_offset_bounds): Same.
8277 * tree-diagnostic.c (default_tree_printer): Move usage of
8278 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
8280 * tree-pretty-print.c (percent_K_format): Add argument.
8281 * tree-pretty-print.h: Add argument.
8282 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
8283 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
8284 (maybe_diag_stxncpy_trunc): Same.
8285 (handle_builtin_stxncpy): Same.
8286 (handle_builtin_strcat): Same.
8288 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8290 * match.pd: Optimise pointer range checks.
8292 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8294 PR tree-optimization/86758
8295 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
8296 to remove pattern statements.
8298 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8300 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
8301 result of dfs_enumerate_from when constructing stmt_vec_infos,
8302 instead of additionally calling get_loop_body.
8304 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8306 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
8308 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
8309 When creating an iv, assert that the step is not known to be zero.
8310 (vect_setup_realignment): Update call accordingly.
8311 * tree-vect-stmts.c (vectorizable_store): Likewise.
8312 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
8314 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8316 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
8317 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
8318 (vectorizable_reduction): Likewise.
8319 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
8320 (vect_detect_hybrid_slp_stmts): Likewise.
8321 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
8323 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
8325 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
8326 (wide_int_set_zero_nonzero_bits): ...this.
8327 (zero_nonzero_bits_from_vr): Rename to...
8328 (vrp_set_zero_nonzero_bits): ...this.
8329 (extract_range_from_multiplicative_op_1): Abstract wide int
8331 (wide_int_range_multiplicative_op): ...here.
8332 (extract_range_from_binary_expr_1): Extract wide int binary
8333 operations into their own functions.
8334 (wide_int_range_lshift): New.
8335 (wide_int_range_can_optimize_bit_op): New.
8336 (wide_int_range_shift_undefined_p): New.
8337 (wide_int_range_bit_xor): New.
8338 (wide_int_range_bit_ior): New.
8339 (wide_int_range_bit_and): New.
8340 (wide_int_range_trunc_mod): New.
8341 (extract_range_into_wide_ints): New.
8342 (vrp_shift_undefined_p): New.
8343 (extract_range_from_multiplicative_op): New.
8344 (vrp_can_optimize_bit_op): New.
8345 * tree-vrp.h (value_range::dump): New.
8346 (wide_int_range_multiplicative_op): New.
8347 (wide_int_range_lshift):New.
8348 (wide_int_range_shift_undefined_p): New.
8349 (wide_int_range_bit_xor): New.
8350 (wide_int_range_bit_ior): New.
8351 (wide_int_range_bit_and): New.
8352 (wide_int_range_trunc_mod): New.
8353 (zero_nonzero_bits_from_bounds): Rename to...
8354 (wide_int_set_zero_nonzero_bits): ...this.
8355 (zero_nonzero_bits_from_vr): Rename to...
8356 (vrp_set_zero_nonzero_bits): ...this.
8357 (range_easy_mask_min_max): Rename to...
8358 (wide_int_range_can_optimize_bit_op): this.
8359 * vr-values.c (simplify_bit_ops_using_ranges): Rename
8360 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
8362 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8364 * tree-vectorizer.h (vect_orig_stmt): New function.
8365 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
8366 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
8367 (vect_create_epilog_for_reduction): Likewise.
8368 (vectorizable_live_operation): Likewise.
8369 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
8370 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
8371 * tree-vect-stmts.c (vectorizable_call): Likewise.
8372 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
8374 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8376 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
8378 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
8379 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
8380 (vect_transform_loop): Likewise.
8381 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
8383 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8385 * tree-vectorizer.h (vect_schedule_slp): Return void.
8386 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
8387 (vect_schedule_slp): Likewise.
8389 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8391 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
8393 (vect_transform_loop): Update calls accordingly. Schedule SLP
8394 instances before the main loop, if any exist.
8396 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8398 PR tree-optimization/86749
8399 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
8400 If the lhs is used in a COND_EXPR, check that it is being used
8401 as the "then" or "else" value.
8403 2018-08-01 Tom de Vries <tdevries@suse.de>
8406 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
8407 speculation_safe_value_not_needed.
8409 2018-08-01 Richard Biener <rguenther@suse.de>
8411 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
8414 2018-08-01 Martin Liska <mliska@suse.cz>
8416 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
8417 format how successful transformation is dumped.
8418 (gimple_mod_pow2_value_transform): Likewise.
8419 (gimple_mod_subtract_transform): Likewise.
8420 (gimple_stringops_transform): Likewise.
8422 2018-08-01 Martin Liska <mliska@suse.cz>
8425 * value-prof.c (interesting_stringop_to_profile_p):
8426 Simplify the code and add BUILT_IN_MEMMOVE.
8427 (gimple_stringops_transform): Likewise.
8429 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
8431 * config/aarch64/aarch64-simd.md
8432 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
8433 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
8434 use GPI iterator instead of SI mode.
8436 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
8438 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
8439 rs6000_speculation_barrier.
8440 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
8441 new barrier pattern name.
8443 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
8445 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
8446 (speculation_barrier): New insn.
8448 2018-08-01 Richard Biener <rguenther@suse.de>
8451 * graphite.h: Include isl/id.h and isl/space.h to allow build
8454 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
8457 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
8458 mode for COFF targets.
8459 * defaults.h (TARGET_COFF): Define.
8460 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
8461 TARGET_COFF): Define.
8462 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
8463 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
8464 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
8466 2018-07-31 Alexandre Oliva <oliva@adacore.com>
8468 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
8469 * gimple-streamer-out.c (output_bb): Save it.
8470 * lto-streamer-in.c (input_struct_function_base): Restore
8471 instance discriminator if available. Create map on demand.
8472 * lto-streamer-out.c (output_struct_function_base): Save it if
8474 * final.c (decl_to_instance_map): Document LTO strategy.
8476 2018-07-31 Alexandre Oliva <oliva@adacore.com>
8477 Olivier Hainque <hainque@adacore.com>
8479 * debug.h (decl_to_instance_map_t): New type.
8480 (decl_to_instance_map): Declare.
8481 (maybe_create_decl_to_instance_map): New inline function.
8482 * final.c (bb_discriminator, last_bb_discriminator): New statics,
8483 to track basic block discriminators.
8484 (final_start_function_1): Initialize them.
8485 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
8487 (decl_to_instance_map): New variable.
8488 (map_decl_to_instance, maybe_set_discriminator): New functions.
8489 (notice_source_line): Set discriminator.
8491 2018-07-31 Ian Lance Taylor <iant@golang.org>
8493 * targhooks.c (default_have_speculation_safe_value): Add
8496 2018-07-31 David Malcolm <dmalcolm@redhat.com>
8498 * dump-context.h: Include "pretty-print.h".
8499 (dump_context::refresh_dumps_are_enabled): New decl.
8500 (dump_context::emit_item): New decl.
8501 (class dump_context): Add fields "m_test_pp" and
8503 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
8504 (temp_dump_context::get_dumped_text): New decl.
8505 (class temp_dump_context): Add field "m_pp".
8506 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
8507 (dump_context::refresh_dumps_are_enabled): ...and add a test for
8509 (set_dump_file): Update for above change.
8510 (set_alt_dump_file): Likewise.
8511 (dump_loc): New overload, taking a pretty_printer *.
8512 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
8513 to any test pretty-printer.
8514 (make_item_for_dump_gimple_stmt): New function, adapted from
8515 optinfo::add_gimple_stmt.
8516 (dump_context::dump_gimple_stmt): Call it, and use the result,
8517 eliminating the direct usage of dump_file and alt_dump_file in
8518 favor of indirectly using them via emit_item.
8519 (make_item_for_dump_gimple_expr): New function, adapted from
8520 optinfo::add_gimple_expr.
8521 (dump_context::dump_gimple_expr): Call it, and use the result,
8522 eliminating the direct usage of dump_file and alt_dump_file in
8523 favor of indirectly using them via emit_item.
8524 (make_item_for_dump_generic_expr): New function, adapted from
8526 (dump_context::dump_generic_expr): Call it, and use the result,
8527 eliminating the direct usage of dump_file and alt_dump_file in
8528 favor of indirectly using them via emit_item.
8529 (make_item_for_dump_printf_va): New function, adapted from
8530 optinfo::add_printf_va.
8531 (make_item_for_dump_printf): New function.
8532 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
8533 and use the result, eliminating the direct usage of dump_file and
8534 alt_dump_file in favor of indirectly using them via emit_item.
8535 (make_item_for_dump_dec): New function.
8536 (dump_context::dump_dec): Call it, and use the result,
8537 eliminating the direct usage of dump_file and alt_dump_file in
8538 favor of indirectly using them via emit_item.
8539 (make_item_for_dump_symtab_node): New function, adapted from
8540 optinfo::add_symtab_node.
8541 (dump_context::dump_symtab_node): Call it, and use the result,
8542 eliminating the direct usage of dump_file and alt_dump_file in
8543 favor of indirectly using them via emit_item.
8544 (dump_context::begin_scope): Reimplement, avoiding direct usage
8545 of dump_file and alt_dump_file in favor of indirectly using them
8547 (dump_context::emit_item): New member function.
8548 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
8549 Set up test pretty-printer on the underlying context. Call
8550 refresh_dumps_are_enabled.
8551 (temp_dump_context::~temp_dump_context): Call
8552 refresh_dumps_are_enabled.
8553 (temp_dump_context::get_dumped_text): New member function.
8554 (selftest::verify_dumped_text): New function.
8555 (ASSERT_DUMPED_TEXT_EQ): New macro.
8556 (selftest::test_capture_of_dump_calls): Run all tests twice, with
8557 and then without optinfo enabled. Add uses of
8558 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
8559 * dumpfile.h: Update comment for the dump_* API.
8560 * optinfo-emit-json.cc
8561 (selftest::test_building_json_from_dump_calls): Update for new
8562 param for temp_dump_context ctor.
8563 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
8564 and "m_owned" field.
8565 (optinfo_item::~optinfo_item): Likewise.
8566 (optinfo::add_item): New member function.
8567 (optinfo::emit): Update comment.
8568 (optinfo::add_string): Delete.
8569 (optinfo::add_printf): Delete.
8570 (optinfo::add_printf_va): Delete.
8571 (optinfo::add_gimple_stmt): Delete.
8572 (optinfo::add_gimple_expr): Delete.
8573 (optinfo::add_tree): Delete.
8574 (optinfo::add_symtab_node): Delete.
8575 (optinfo::add_dec): Delete.
8576 * optinfo.h (class dump_context): New forward decl.
8577 (optinfo::add_item): New decl.
8578 (optinfo::add_string): Delete.
8579 (optinfo::add_printf): Delete.
8580 (optinfo::add_printf_va): Delete.
8581 (optinfo::add_gimple_stmt): Delete.
8582 (optinfo::add_gimple_expr): Delete.
8583 (optinfo::add_tree): Delete.
8584 (optinfo::add_symtab_node): Delete.
8585 (optinfo::add_dec): Delete.
8586 (optinfo::add_poly_int): Delete.
8587 (optinfo_item::optinfo_item): Remove "owned" param.
8588 (class optinfo_item): Remove field "m_owned".
8590 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
8593 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
8594 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
8595 requested variable alignment.
8596 (expand_one_ssa_partition): Likewise.
8597 (expand_one_var): Likewise.
8599 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8601 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
8602 to speculation_safe_value_not_needed.
8604 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8606 * targhooks.h (speculation_safe_value_not_needed): New prototype.
8607 * targhooks.c (speculation_safe_value_not_needed): New function.
8608 * target.def (have_speculation_safe_value): Update documentation.
8609 * doc/tm.texi: Regenerated.
8611 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8613 * config/aarch64/iterators.md (ALLI_TI): New iterator.
8614 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
8616 (despeculate_copy<ALLI:mode>_insn): New insn.
8617 (despeculate_copyti_insn): New insn.
8618 (despeculate_simple<ALLI:mode>): New insn
8619 (despeculate_simpleti): New insn.
8620 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
8622 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
8623 aarch64_speculation_safe_value.
8624 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
8626 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8628 * config/aarch64/aarch64-speculation.cc: New file.
8629 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
8630 before pass_reorder_blocks.
8631 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
8633 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
8634 X14 and X15 when tracking speculation.
8635 * config/aarch64/aarch64.md (register name constants): Add
8636 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
8637 (unspec): Add UNSPEC_SPECULATION_TRACKER.
8638 (speculation_barrier): New insn attribute.
8639 (cmp<mode>): Allow SP in comparisons.
8640 (speculation_tracker): New insn.
8641 (speculation_barrier): Add speculation_barrier attribute.
8642 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
8643 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
8644 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
8646 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8648 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
8649 aarch64_track_speculation is true.
8650 (tb<optab><mode>1): Likewise.
8651 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
8652 generate CB[N]Z when tracking speculation.
8653 (aarch64_split_compare_and_swap): Likewise.
8654 (aarch64_split_atomic_op): Likewise.
8656 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8658 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
8660 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8662 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
8663 (speculation_barrier): New insn.
8665 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8667 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
8668 * config/arm/arm.md (speculation_barrier): New expand.
8669 (speculation_barrier_insn): New pattern.
8671 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8673 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
8674 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
8675 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
8676 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
8678 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
8679 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
8680 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
8681 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
8682 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
8683 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
8684 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
8685 * builtins.c (expand_speculation_safe_value): New function.
8686 (expand_builtin): Call it.
8687 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
8688 * doc/extend.texi: Document __builtin_speculation_safe_value.
8689 * doc/md.texi: Document "speculation_barrier" pattern.
8690 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
8691 TARGET_HAVE_SPECULATION_SAFE_VALUE.
8692 * doc/tm.texi: Regenerated.
8693 * target.def (have_speculation_safe_value, speculation_safe_value): New
8695 * targhooks.c (default_have_speculation_safe_value): New function.
8696 (default_speculation_safe_value): New function.
8697 * targhooks.h (default_have_speculation_safe_value): Add prototype.
8698 (default_speculation_safe_value): Add prototype.
8700 2018-07-31 David Malcolm <dmalcolm@redhat.com>
8702 * dump-context.h (dump_context::dump_loc): New decl.
8703 * dumpfile.c (dump_context::dump_loc): New member function.
8704 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
8705 and dump_gimple_stmt.
8706 (dump_context::dump_gimple_expr_loc): Likewise, using
8708 (dump_context::dump_generic_expr_loc): Likewise, using
8710 (dump_context::dump_printf_loc_va): Likewise, using
8712 (dump_context::begin_scope): Explicitly using the global function
8713 "dump_loc", rather than the member function.
8715 2018-07-31 Martin Sebor <msebor@redhat.com>
8717 PR tree-optimization/86741
8718 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
8720 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
8722 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
8723 depend on whether prefetch instructions will be emitted or not.
8724 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
8725 will be emitted or not.
8726 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
8727 (TARGET_SETMEM_PFD): New macros.
8729 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8731 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
8732 (NULL_STMT_VEC_INFO): Delete.
8733 (stmt_vec_info::operator*): Likewise.
8734 (stmt_vec_info::operator gimple *): Likewise.
8735 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
8736 of NULL_STMT_VEC_INFO.
8737 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
8738 (vect_reassociating_reduction_p): Likewise.
8739 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
8740 (vectorizable_store): Likewise.
8741 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
8742 (vec_info::free_stmt_vec_infos): Likewise.
8744 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8746 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
8747 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
8749 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8751 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
8752 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
8753 (vec_info::free_stmt_vec_info): New private member functions.
8754 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
8755 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
8756 * tree-parloops.c (gather_scalar_reductions): Remove calls to
8757 set_stmt_vec_info_vec and free_stmt_vec_infos.
8758 * tree-vect-loop.c (_loop_vec_info): Remove call to
8759 set_stmt_vec_info_vec.
8760 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
8761 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
8762 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
8763 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
8764 (vec_info::free_stmt_vec_info): ...these new functions. Remove
8765 assignments in {vec_info::,}new_stmt_vec_info that are redundant
8766 with the clearing in the xcalloc.
8767 (stmt_vec_info_vec): Delete.
8768 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
8769 (vectorize_loops): Likewise.
8770 (vec_info::~vec_info): Remove argument from call to
8771 free_stmt_vec_infos.
8772 (vec_info::add_stmt): Remove vinfo argument from call to
8775 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8777 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
8778 rather than a gimple stmt.
8779 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
8780 information for pattern statements when passed the original
8781 statement; instead wait to be passed the pattern statement itself.
8782 Don't call set_vinfo_for_stmt here.
8783 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
8784 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
8785 stmt_vec_infos here.
8786 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
8787 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
8788 stmt_vec_infos entry.
8790 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8792 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
8793 * tree-vectorizer.c (vec_info::replace_stmt): New function.
8794 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
8795 * tree-vect-stmts.c (vectorizable_call): Likewise.
8796 (vectorizable_simd_clone_call): Likewise.
8798 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8800 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
8801 * tree-vectorizer.c (vec_info::remove_stmt): New function.
8802 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
8803 * tree-vect-loop.c (vect_transform_loop): Likewise.
8804 * tree-vect-slp.c (vect_schedule_slp): Likewise.
8805 * tree-vect-stmts.c (vect_remove_stores): Likewise.
8807 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8809 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
8810 (vect_dr_stmt): Delete.
8811 * tree-vectorizer.c (vec_info::lookup_dr): New function.
8812 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
8814 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
8815 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
8816 (vect_verify_datarefs_alignment, vect_peeling_supportable)
8817 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
8818 (vect_analyze_data_refs): Likewise.
8819 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
8821 (vect_find_same_alignment_drs): Likewise.
8822 (vect_slp_analyze_node_dependences): Update calls accordingly.
8823 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
8824 instead of DR_VECT_AUX.
8825 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
8826 of a vector data references. Use vec_info::lookup_dr instead of
8828 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
8829 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
8830 instead of DR_VECT_AUX.
8832 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8834 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
8836 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
8838 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
8839 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
8840 (vect_gen_prolog_loop_niters): Likewise.
8842 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8844 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
8845 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
8846 (vect_known_alignment_in_bytes, vect_dr_behavior)
8847 (vect_get_scalar_dr_size): Take references as dr_vec_infos
8848 instead of data_references. Update calls to other routines for
8849 which the same change has been made.
8850 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
8851 dr_vec_infos instead of stmt_vec_infos.
8852 (vect_analyze_data_ref_dependence): Update call accordingly.
8853 (vect_slp_analyze_data_ref_dependence)
8854 (vect_record_base_alignments): Use DR_VECT_AUX.
8855 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
8856 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
8857 (vector_alignment_reachable_p, vect_get_data_access_cost)
8858 (vect_peeling_supportable, vect_analyze_group_access_1)
8859 (vect_analyze_group_access, vect_analyze_data_ref_access)
8860 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
8861 (vect_compile_time_alias, vect_small_gap_p)
8862 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
8863 (vect_supportable_dr_alignment): Take references as dr_vec_infos
8864 instead of data_references. Update calls to other routines for
8865 which the same change has been made.
8866 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
8867 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
8868 (vect_slp_analyze_and_verify_node_alignment)
8869 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
8870 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
8871 (vect_setup_realignment): Use dr_vec_infos. Update calls after
8873 (_vect_peel_info::dr): Replace with...
8874 (_vect_peel_info::dr_info): ...this new field.
8875 (vect_peeling_hash_get_most_frequent)
8876 (vect_peeling_hash_choose_best_peeling): Update accordingly.
8877 (vect_peeling_hash_get_lowest_cost):
8878 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
8879 routines for which the same change has been made.
8880 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
8882 * tree-vect-loop-manip.c (get_misalign_in_elems)
8883 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
8885 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
8886 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
8887 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
8888 (get_group_load_store_type, get_negative_load_store_type)
8889 (vect_get_data_ptr_increment, vectorizable_store)
8890 (vectorizable_load): Likewise.
8891 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
8892 Update calls to other routines for which the same change has been made.
8894 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8896 * tree-vectorizer.h (vec_info::move_dr): New member function.
8897 (dataref_aux): Rename to...
8898 (dr_vec_info): ...this and add "dr" and "stmt" fields.
8899 (_stmt_vec_info::dr_aux): Update accordingly.
8900 (_stmt_vec_info::data_ref_info): Delete.
8901 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
8902 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
8903 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
8904 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
8906 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
8907 (STMT_VINFO_DR_INFO): New macro.
8908 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
8909 (set_dr_misalignment): Update after rename of dataref_aux.
8910 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
8911 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
8912 initialization of STMT_VINFO_DATA_REF.
8913 * tree-vectorizer.c (vec_info::move_dr): New function.
8914 * tree-vect-patterns.c (vect_recog_bool_pattern)
8915 (vect_recog_mask_conversion_pattern)
8916 (vect_recog_gather_scatter_pattern): Use it.
8917 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
8918 the "dr" and "stmt" fields of dr_vec_info instead of
8919 STMT_VINFO_DATA_REF.
8921 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8923 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
8924 (is_pattern_stmt_p): Use it.
8925 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
8926 on pattern statements.
8928 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8930 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
8931 original stmt as a stmt_vec_info rather than a gimple stmt.
8932 (vect_pattern_recog_1): Take the statement directly as a
8933 stmt_vec_info, rather than via a gimple_stmt_iterator.
8934 Update call to vect_mark_pattern_stmts.
8935 (vect_pattern_recog): Update calls accordingly.
8937 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8939 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
8940 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
8941 a vect_def_type for the first argument.
8942 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
8943 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
8944 operand if it isn't defined by a vectorized statement.
8945 (vect_build_gather_load_calls): Remove the mask_dt argument and
8946 update calls to vect_get_vec_def_for_stmt_copy.
8947 (vectorizable_bswap): Likewise the dt argument.
8948 (vectorizable_call): Update calls to vectorizable_bswap and
8949 vect_get_vec_def_for_stmt_copy.
8950 (vectorizable_simd_clone_call, vectorizable_assignment)
8951 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
8952 (vectorizable_comparison): Update calls to
8953 vect_get_vec_def_for_stmt_copy.
8954 (vectorizable_store): Likewise. Remove now-unnecessary calls to
8956 (vect_get_loop_based_defs): Remove dt argument and update call
8957 to vect_get_vec_def_for_stmt_copy.
8958 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
8959 and vect_get_vec_def_for_stmt_copy.
8960 (vectorizable_load): Update calls to vect_build_gather_load_calls
8961 and vect_get_vec_def_for_stmt_copy.
8962 * tree-vect-loop.c (vect_create_epilog_for_reduction)
8963 (vectorizable_reduction, vectorizable_live_operation): Update calls
8964 to vect_get_vec_def_for_stmt_copy.
8966 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8968 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
8969 and gimple stmt arguments with a stmt_vec_info.
8970 (vect_record_base_alignments): Update calls accordingly.
8971 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
8972 and gimple stmt arguments with a stmt_vec_info.
8973 (vect_build_slp_tree_1): Remove vinfo argument and update call
8974 to vect_record_max_nunits.
8975 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
8976 and vect_record_max_nunits.
8978 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8980 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
8981 file and take a stmt_vec_info instead of a gimple stmt.
8982 (supportable_widening_operation, vect_finish_replace_stmt)
8983 (vect_finish_stmt_generation, vect_get_store_rhs)
8984 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
8985 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
8986 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
8987 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
8988 (vect_create_data_ref_ptr, bump_vector_ptr)
8989 (vect_permute_store_chain, vect_setup_realignment)
8990 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
8991 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
8992 (vectorizable_reduction, vectorizable_induction)
8993 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
8994 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
8995 than gimple stmts as arguments.
8996 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
8997 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
8998 (can_group_stmts_p, vect_check_gather_scatter)
8999 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
9000 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
9001 (vect_permute_load_chain, vect_shift_permute_load_chain)
9002 (vect_transform_grouped_load)
9003 (vect_record_grouped_load_vectors): Likewise.
9004 * tree-vect-loop.c (vect_fixup_reduc_chain)
9005 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
9006 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
9007 (vectorizable_reduction, vectorizable_induction)
9008 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
9009 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
9010 (vect_get_load_store_mask): Likewise.
9011 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
9012 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
9013 * tree-vect-stmts.c (vect_mark_relevant)
9014 (is_simple_and_all_uses_invariant)
9015 (exist_non_indexing_operands_for_use_p, process_use)
9016 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
9017 (vect_get_vec_def_for_operand, vect_get_vec_defs)
9018 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
9019 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
9020 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
9021 (get_negative_load_store_type, get_load_store_type)
9022 (vect_check_load_store_mask, vect_check_store_rhs)
9023 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
9024 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
9025 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
9026 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
9027 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
9028 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
9029 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
9030 (supportable_widening_operation): Likewise.
9032 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9034 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
9035 a stmt_vec_info instead of a gcall.
9036 (vect_check_gather_scatter): Update call accordingly.
9037 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
9039 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
9040 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
9041 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
9042 instead of a gimple stmt.
9043 (vect_transform_loop): Update calls accordingly.
9044 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
9045 stmt_vec_infos instead of gimple stmts.
9046 (vect_analyze_slp_instance): Update use accordingly.
9047 * tree-vect-stmts.c (read_vector_array, write_vector_array)
9048 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
9049 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
9050 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
9051 (vect_gen_widened_results_half, vect_get_loop_based_defs)
9052 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
9053 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
9054 down to subroutines.
9056 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9058 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
9059 of the worklist from a vector of gimple stmts to a vector of
9061 * tree-vect-stmts.c (vect_mark_relevant, process_use)
9062 (vect_mark_stmts_to_be_vectorized): Likewise
9064 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9066 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
9067 statement before passing it to vect_analyze_stmt.
9068 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
9069 the chain of phi vector definitions. Track the exit phi via its
9071 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
9072 STMT_VINFO_REDUC_DEF.
9073 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
9074 stmt_vec_infos to handle the statement chains.
9075 (vect_get_slp_defs): Record the first statement in the node
9076 using a stmt_vec_info.
9077 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
9078 statements here and pass their stmt_vec_info down to subroutines.
9079 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
9080 down to vect_finish_stmt_generation.
9081 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
9082 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
9083 stmt_vec_infos to subroutines.
9084 (vect_remove_stores): Use stmt_vec_infos to handle the statement
9087 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9089 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
9090 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
9091 (vect_permute_store_chain, vect_setup_realignment)
9092 (vect_permute_load_chain, vect_shift_permute_load_chain)
9093 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
9094 stmts internally, and when passing values to other vectorizer routines.
9095 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
9096 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
9097 (vect_analyze_loop_operations, get_initial_def_for_reduction)
9098 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
9099 (vectorizable_reduction, vectorizable_induction)
9100 (vectorizable_live_operation, vect_transform_loop_stmt)
9101 (vect_transform_loop): Likewise.
9102 * tree-vect-patterns.c (vect_reassociating_reduction_p)
9103 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
9104 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
9105 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
9106 (vect_slp_analyze_node_operations_1): Likewise.
9107 * tree-vect-stmts.c (vect_mark_relevant, process_use)
9108 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
9109 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
9110 (vect_finish_stmt_generation_1, get_group_load_store_type)
9111 (get_load_store_type, vect_build_gather_load_calls)
9112 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
9113 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
9114 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
9115 (vectorizable_store, vectorizable_load, vectorizable_condition)
9116 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
9117 (supportable_widening_operation): Likewise.
9118 (vect_get_vector_types_for_stmt): Likewise.
9119 * tree-vectorizer.h (vect_dr_behavior): Likewise.
9121 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9123 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
9124 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
9125 (vect_permute_store_chain, vect_permute_load_chain)
9126 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
9127 repeated stmt_vec_info lookups.
9128 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
9129 (vect_update_ivs_after_vectorizer): Likewise.
9130 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
9131 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
9132 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
9133 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
9134 (vect_bb_slp_scalar_cost): Likewise.
9135 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
9137 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9139 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
9140 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
9141 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
9142 to get gassigns and gcalls, rather than operating on generc gimple
9144 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
9145 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
9146 (vectorizable_load, vect_analyze_stmt): Likewise.
9147 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
9149 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9151 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
9152 return stmt_vec_infos rather than gimple stmts. Do not accept
9154 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
9156 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
9157 Update use of get_later_stmt.
9158 (vect_get_constant_vectors): Update call accordingly.
9159 (vect_schedule_slp_instance): Likewise
9160 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
9161 (vect_slp_analyze_instance_dependence): Likewise.
9162 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
9164 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9166 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
9167 (stmt_info_for_cost::stmt_info): ...this new field.
9168 (add_stmt_costs): Update accordingly.
9169 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
9170 (vect_get_known_peeling_cost): Likewise.
9171 (vect_estimate_min_profitable_iters): Likewise.
9172 * tree-vect-stmts.c (record_stmt_cost): Likewise.
9174 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9176 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
9177 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
9178 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
9180 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
9182 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9184 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
9185 a gimple stmt to a stmt_vec_info.
9186 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
9188 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9190 * tree-vectorizer.h (vec_info::grouped_stores): Change from
9191 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
9192 (_loop_vec_info::reduction_chains): Likewise.
9193 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
9195 * tree-vect-slp.c (vect_analyze_slp): Likewise.
9197 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9199 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
9200 a gimple stmt to a stmt_vec_info.
9201 (_stmt_vec_info::next_element): Likewise.
9202 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
9203 (vect_slp_analyze_and_verify_node_alignment)
9204 (vect_analyze_group_access_1, vect_analyze_group_access)
9205 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
9206 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
9207 (vect_supportable_dr_alignment): Update accordingly.
9208 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
9209 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
9210 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
9211 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
9212 * tree-vect-slp.c (vect_build_slp_tree_1)
9213 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
9214 (vect_split_slp_store_group, vect_analyze_slp_instance)
9215 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
9216 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
9217 (get_group_load_store_type, get_load_store_type)
9218 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
9219 (vect_transform_stmt, vect_remove_stores): Likewise.
9221 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9223 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
9225 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
9226 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
9227 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
9228 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
9229 (vector_alignment_reachable_p, vect_get_data_access_cost)
9230 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
9231 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
9232 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
9233 (vect_analyze_group_access_1, vect_analyze_group_access)
9234 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
9235 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
9236 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
9237 result of vect_dr_stmt and use the stmt_vec_info instead of
9238 the associated gimple stmt.
9239 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
9240 (vect_gen_prolog_loop_niters): Likewise.
9241 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
9243 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9245 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
9246 to a vec<stmt_vec_info>.
9247 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
9248 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
9250 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
9251 to a vec<stmt_vec_info>.
9252 (bst_traits::value_type, bst_traits::value_type): Likewise.
9253 (bst_traits::hash): Update accordingly.
9254 (vect_get_and_check_slp_defs): Change the stmts parameter from
9255 a vec<gimple *> to a vec<stmt_vec_info>.
9256 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
9257 (vect_build_slp_tree): Likewise.
9258 (vect_build_slp_tree_2): Likewise. Update uses of
9259 SLP_TREE_SCALAR_STMTS.
9260 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
9261 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
9262 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
9263 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
9264 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
9265 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
9266 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
9267 (vect_get_constant_vectors, vect_get_slp_defs)
9268 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
9269 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
9270 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
9271 instead of gimple stmts.
9272 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
9273 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
9274 (vect_slp_analyze_instance_dependence): Update uses of
9275 SLP_TREE_SCALAR_STMTS.
9276 (vect_slp_analyze_and_verify_node_alignment): Likewise.
9277 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
9278 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
9279 (get_initial_defs_for_reduction): Likewise.
9280 (vect_create_epilog_for_reduction): Likewise.
9281 (vectorize_fold_left_reduction): Likewise.
9282 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
9283 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
9284 (can_vectorize_live_stmts): Likewise.
9286 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9288 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
9289 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
9290 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
9292 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
9294 (gather_scalar_reductions): Update after above interface changes.
9295 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
9296 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
9298 (vect_force_simple_reduction): Likewise.
9299 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
9300 LOOP_VINFO_REDUCTIONS.
9301 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
9303 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9305 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
9306 a gimple stmt to a stmt_vec_info.
9307 * tree-vect-loop.c (vect_active_double_reduction_p)
9308 (vect_force_simple_reduction, vectorizable_reduction): Update
9311 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9313 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
9314 vec<gimple *> to a vec<stmt_vec_info>.
9315 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
9316 the reduction_phis argument from a vec<gimple *> to a
9318 (vectorizable_reduction): Likewise the phis local variable that
9319 is passed to vect_create_epilog_for_reduction. Update for new type
9320 of SLP_TREE_VEC_STMTS.
9321 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
9322 (vectorizable_live_operation): Likewise.
9323 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
9324 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
9326 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9328 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
9329 a gimple stmt to a stmt_vec_info.
9330 (vectorizable_condition, vectorizable_live_operation)
9331 (vectorizable_reduction, vectorizable_induction): Pass back the
9332 vectorized statement as a stmt_vec_info.
9333 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
9334 use of STMT_VINFO_VEC_STMT.
9335 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
9336 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
9337 as stmt_vec_infos rather than gimple stmts.
9338 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
9340 (vectorizable_live_operation): Likewise.
9341 (vectorizable_reduction, vectorizable_induction): Likewise,
9342 updating use of STMT_VINFO_VEC_STMT.
9343 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
9344 of STMT_VINFO_VEC_STMT.
9345 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
9346 (vectorizable_simd_clone_call, vectorizable_conversion)
9347 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
9348 (vectorizable_store, vectorizable_load, vectorizable_condition)
9349 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
9350 from a gimple stmt to a stmt_vec_info.
9351 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
9352 pointer to a stmt_vec_info to the vectorizable_* routines.
9354 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9356 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
9357 a gimple stmt to a stmt_vec_info.
9358 (is_pattern_stmt_p): Update accordingly.
9359 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
9360 (vect_record_grouped_load_vectors): Likewise.
9361 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
9362 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
9363 (vect_model_reduction_cost): Likewise.
9364 (vect_create_epilog_for_reduction): Likewise.
9365 (vectorizable_reduction, vectorizable_induction): Likewise.
9366 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
9367 Return the stmt_vec_info for the pattern statement.
9368 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
9369 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
9370 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
9371 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
9372 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
9373 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
9374 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
9375 (free_stmt_vec_info, vect_is_simple_use): Likewise.
9377 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9379 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
9380 (vect_finish_stmt_generation): Likewise.
9381 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
9382 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
9383 (vect_build_gather_load_calls): Use the return value of the above
9384 functions instead of a separate call to vinfo_for_stmt. Use narrow
9385 scopes for the input gimple stmt and wider scopes for the associated
9386 stmt_vec_info. Use vec_info::lookup_def when setting these
9387 stmt_vec_infos from an SSA_NAME definition.
9388 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
9389 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
9390 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
9391 (vectorizable_store, vectorizable_load, vectorizable_condition)
9392 (vectorizable_comparison): Likewise.
9393 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
9394 (vectorizable_reduction): Likewise.
9396 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9398 * tree-vectorizer.h (vect_is_simple_use): Add an optional
9399 stmt_vec_info * parameter before the optional gimple **.
9400 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
9401 (process_use, vect_get_vec_def_for_operand_1): Update callers.
9402 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
9403 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9404 (vectorizable_live_operation): Likewise.
9405 * tree-vect-patterns.c (type_conversion_p): Likewise.
9406 (vect_look_through_possible_promotion): Likewise.
9407 (vect_recog_rotate_pattern): Likewise.
9408 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
9410 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9412 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
9413 a typedef to a wrapper class.
9414 (NULL_STMT_VEC_INFO): New macro.
9415 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
9416 (stmt_vec_info::operator*): New function.
9417 (stmt_vec_info::operator gimple *): Likewise.
9418 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
9419 (add_stmt_costs): Likewise.
9420 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
9421 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
9422 (vect_get_known_peeling_cost): Likewise.
9423 (vect_estimate_min_profitable_iters): Likewise.
9424 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
9425 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
9426 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
9427 (vectorizable_store, free_stmt_vec_infos): Likewise.
9428 (new_stmt_vec_info): Change return type of xcalloc to
9431 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9433 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
9434 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
9435 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
9436 a single_imm_use-based sequence.
9437 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
9439 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9441 * tree-vectorizer.h (vec_info::lookup_def): Declare.
9442 * tree-vectorizer.c (vec_info::lookup_def): New function.
9443 * tree-vect-patterns.c (vect_get_internal_def): Use it.
9444 (vect_widened_op_tree): Likewise.
9445 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
9446 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
9447 (vectorizable_reduction): Likewise.
9448 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
9450 (vect_is_slp_reduction): Update calls accordingly. Use
9451 vec_info::lookup_def.
9452 (vect_is_simple_reduction): Likewise
9453 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
9455 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9457 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
9458 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
9459 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
9461 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
9462 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
9463 (vect_update_vf_for_slp, vect_analyze_loop_operations)
9464 (vect_is_slp_reduction, vectorizable_induction)
9465 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
9466 * tree-vect-patterns.c (vect_init_pattern_stmt):
9467 (vect_determine_min_output_precision_1, vect_determine_precisions)
9468 (vect_pattern_recog): Likewise.
9469 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
9470 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
9471 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
9472 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
9473 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
9474 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
9475 info field from a loop to a loop_vec_info.
9477 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9479 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
9480 (vec_info::add_stmt): Declare.
9481 * tree-vectorizer.c (vec_info::add_stmt): New function.
9482 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
9483 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
9484 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
9485 (vectorizable_induction): Likewise.
9486 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
9487 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
9488 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
9489 (vectorizable_load): Likewise.
9490 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
9491 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
9492 (vect_recog_gather_scatter_pattern): Likewise.
9493 (append_pattern_def_seq): Likewise. Remove a check that is
9494 performed by add_stmt itself.
9496 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9498 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
9499 which make_ssa_name was called with new_stmt before new_stmt
9502 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9504 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
9506 (vect_is_slp_reduction): ...here...
9507 (vect_is_simple_reduction): ...and here. Remove repetition of tests
9508 that are already known to be false.
9510 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9512 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
9513 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
9514 STMT_VINFO_NUM_SLP_USES when it's true.
9515 (vect_free_slp_instance): Add a final_p parameter and pass it to
9517 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
9518 (vect_analyze_slp_instance): Likewise.
9519 (vect_slp_analyze_operations): Likewise.
9520 (vect_slp_analyze_bb_1): Likewise.
9521 * tree-vectorizer.c (vec_info): Likewise.
9522 * tree-vect-loop.c (vect_transform_loop): Likewise.
9524 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9526 * tree-vect-loop.c (vectorizable_reduction): Assert that the
9527 function is not called for second and subsequent members of
9530 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9532 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
9533 cases for nested loops from here to ...
9534 (vect_create_epilog_for_reduction): ...here. Only call
9535 vect_is_simple_use for inner-loop reductions.
9537 2018-07-31 Martin Liska <mliska@suse.cz>
9539 PR gcov-profile/85338
9540 PR gcov-profile/85350
9541 PR gcov-profile/85372
9542 * profile.c (struct location_triplet): New.
9543 (struct location_triplet_hash): Likewise.
9544 (output_location): Do not output a BB that
9545 is already recorded for a line.
9546 (branch_prob): Use streamed_locations.
9548 2018-07-31 Martin Liska <mliska@suse.cz>
9550 PR gcov-profile/85370
9551 * coverage.c (coverage_begin_function): Do not mark target
9552 clones as artificial functions.
9554 2018-07-31 Martin Liska <mliska@suse.cz>
9556 PR gcov-profile/83813
9557 PR gcov-profile/84758
9558 PR gcov-profile/85217
9559 PR gcov-profile/85332
9560 * profile.c (branch_prob): Do not record GOTO expressions
9561 for GIMPLE statements which locations are already streamed.
9563 2018-07-31 Olivier Hainque <hainque@adacore.com>
9565 * gcc.c (handle_spec_function): Accept a soft_matched_part
9566 argument, as do_spec_1. Pass it down to ...
9567 (eval_spec_function): Accept a soft_matched_part argument,
9568 and pass it down to ...
9569 (do_spec_2): Accept a soft_matched_part argument, and pass
9570 it down to do_spec_1.
9571 (do_spec_1): Pass soft_matched_part to handle_spec_function.
9572 (handle_braces): Update call to handle_spec_function.
9573 (driver::set_up_specs): Update calls to do_spec_2.
9574 (compare_debug_dump_opt_spec_function): Likewise.
9575 (compare_debug_self_opt_spec_function): Likewise.
9577 2018-07-31 Olivier Hainque <hainque@adacore.com>
9579 * common.opt (nolibc): New option.
9580 * doc/invoke.texi (Link Options): Document it.
9581 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
9582 * config/alpha/linux.h: Likewise.
9583 * config/arc/elf.h: Likewise.
9584 * config/arm/uclinux-elf.h: Likewise.
9585 * config/arm/unknown-elf.h: Likewise.
9586 * config/avr/avrlibc.h: Likewise.
9587 * config/bfin/bfin.h: Likewise.
9588 * config/bfin/linux.h: Likewise.
9589 * config/bfin/uclinux.h: Likewise.
9590 * config/darwin.h: Likewise.
9591 * config/darwin10.h: Likewise.
9592 * config/darwin12.h: Likewise.
9593 * config/gnu-user.h: Likewise.
9594 * config/lm32/uclinux-elf.h: Likewise.
9595 * config/pa/pa-hpux11.h: Likewise.
9596 * config/pa/pa64-hpux.h: Likewise.
9597 * config/sparc/sparc.h: Likewise.
9599 2018-07-31 Olivier Hainque <hainque@adacore.com>
9601 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
9602 undefined variables.
9604 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
9607 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
9610 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
9612 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
9613 terminated string literal.
9615 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
9617 PR rtl-optimization/85160
9618 * combine.c (is_just_move): New function.
9619 (try_combine): Allow combining two instructions into two if neither of
9620 the original instructions was a move.
9622 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
9625 * doc/extend.texi (Global Register Variables): Discourage use of type
9627 (Local Register Variables): Likewise.
9629 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
9631 PR tree-optimization/86506
9632 * hwint.h (ceil_log2): Resync with hwint.c implementation.
9634 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
9637 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
9638 hard_regno, make sure no insn between `from` and `to` clobbers it.
9640 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
9641 Tom de Vries <tdevries@suse.de>
9643 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
9644 (PTX_DEFAULT_RUNTIME_DIM): ... this.
9645 (nvptx_goacc_validate_dims): Set default worker and gang dims to
9646 PTX_DEFAULT_RUNTIME_DIM.
9647 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
9649 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
9651 * config/pa/pa.c (pa_output_addr_vec): Align address table.
9652 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
9653 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
9655 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
9657 * config/rs6000/constraints.md (wG constraint): Delete, no longer
9659 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
9660 predicate to reflect toc fusion has been deleted.
9661 (toc_fusion_mem_raw): Delete, no longer used.
9662 (toc_fusion_mem_wrapped): Likewise.
9663 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
9665 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
9666 Delete, no longer used.
9667 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
9668 meant to be used for toc fusion.
9669 (rs6000_debug_print_mode): Delete toc fusion debugging.
9670 (rs6000_debug_reg_global): Likewise.
9671 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
9672 fusion and secondary reload support that were never used.
9673 (rs6000_option_override_internal): Delete TOC fusion, that was only
9674 partially defined, and it did not work unless you also used the
9676 (rs6000_legitimate_address_p): Delete TOC fusion support.
9677 (rs6000_opt_masks): Likewise.
9678 (fusion_wrap_memory_address): Delete function, no longer used.
9679 (fusion_split_address); Delete TOC fusion support.
9680 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
9681 longer used with toc fusion being deleted.
9682 (TARGET_TOC_FUSION_FP): Likewise.
9683 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
9685 (toc fusion spliter): Delete TOC fusion support.
9686 (toc_fusionload_<mode>): Likewise.
9687 (toc_fusionload_di): Likewise.
9688 (fusion_gpr_load_<mode>): Delete generator function, this insn no
9689 longer needs to be named. Rename predicate to delete TOC fusion.
9690 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
9691 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
9692 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
9693 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
9694 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
9696 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
9698 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
9699 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
9700 __int128 in built-in function prototypes.
9701 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
9702 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
9704 2018-07-27 Martin Sebor <msebor@redhat.com>
9706 PR tree-optimization/86696
9707 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
9708 types, including enums.
9709 (handle_char_store): Be prepared for the above function to fail.
9711 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
9713 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
9714 when optimization level is lower than 2 or optimize for size.
9716 2018-07-26 Martin Sebor <msebor@redhat.com>
9718 PR tree-optimization/86043
9719 PR tree-optimization/86042
9720 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
9721 (get_string_cst_length): Rename...
9722 (get_min_string_length): ...to this. Add argument.
9723 (handle_char_store): Extend to handle multi-character stores by
9725 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
9726 * tree.h (initializer_zerop): Add argument.
9728 2018-07-26 Jakub Jelinek <jakub@redhat.com>
9731 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
9732 declare target to variables if they have always,{to,from,tofrom} map
9735 2018-07-26 Martin Liska <mliska@suse.cz>
9738 * lto-wrapper.c: Add linker_output as prefix
9739 for ltrans_output_file.
9741 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
9743 PR rtl-optimization/85805
9744 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
9745 value for hard registers if that was written in the same mode.
9747 2018-07-26 Martin Liska <mliska@suse.cz>
9749 PR gcov-profile/86536
9750 * gcov.c (format_gcov): Use printf format %.*f directly
9751 and do not handle special values.
9753 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
9755 * common/config/arc/arc-common.c (arc_option_optimization_table):
9756 Update default optimizations for size.
9758 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
9760 * config/arc/arc.md (movsf_insn): Add short instruction selection.
9761 * config/arc/constraints.md (CfZ): New constraint.
9762 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
9763 (subsf3_fpu): Likewise.
9764 (cmpsf_fpu): Likewise.
9765 (cmpsf_fpu_uneq): Likewise.
9767 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
9769 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
9771 (arc_is_uncached_mem_p): uncached applies to both the variable and
9774 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
9776 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
9779 2018-07-25 David Malcolm <dmalcolm@redhat.com>
9781 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
9782 field "m_scopes" from vec to auto_vec.
9784 2018-07-25 Martin Liska <mliska@suse.cz>
9786 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
9789 2018-07-25 Richard Biener <rguenther@suse.de>
9792 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
9793 special wrt context_die late.
9794 (gen_subprogram_die): Re-use DIEs in local scope.
9796 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
9798 PR tree-optimization/86644
9799 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
9801 2018-07-25 Martin Liska <mliska@suse.cz>
9804 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
9805 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
9807 2018-07-25 Martin Liska <mliska@suse.cz>
9810 * params.def: Explain ASan abbreviation and provide
9811 a documentation link.
9813 2018-07-24 Martin Sebor <msebor@redhat.com>
9815 PR tree-optimization/86622
9816 PR tree-optimization/86532
9817 * builtins.h (string_length): Declare.
9818 * builtins.c (c_strlen): Correct handling of non-constant offsets.
9819 (check_access): Be prepared for non-constant length ranges.
9820 (string_length): Make extern.
9821 * expr.c (string_constant): Only handle the minor non-constant
9822 array index. Use string_constant to compute the length of
9823 a generic string constant.
9825 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
9827 PR tree-optimization/86618
9828 * tree-vect-stmts.c (vectorizable_call): Don't take the address
9829 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
9831 2018-07-24 David Malcolm <dmalcolm@redhat.com>
9833 PR tree-optimization/86636
9834 * json.cc (json::object::set): Fix comment. Add assertions.
9835 (json::array::append): Move here from json.h. Add comment and an
9837 (json::string::string): Likewise.
9838 * json.h (json::array::append): Move to json.cc.
9839 (json::string::string): Likewise.
9840 * optinfo-emit-json.cc
9841 (optrecord_json_writer::impl_location_to_json): Assert that we
9842 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
9843 wrapper around it. Expand the location once, rather than three
9845 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
9846 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
9848 (optrecord_json_writer::optinfo_to_json): Likewise, in four
9849 places. Fix some overlong lines.
9851 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
9853 * config/aarch64/aarch64-simd.md
9854 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
9855 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
9856 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
9857 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
9858 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
9859 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
9860 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
9861 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
9862 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
9864 2018-07-24 Jakub Jelinek <jakub@redhat.com>
9867 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
9868 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
9869 and abs_d == d, do the power of two handling if profitable.
9871 2018-07-24 Richard Biener <rguenther@suse.de>
9873 * match.pd: Add BIT_FIELD_REF canonicalizations.
9875 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
9878 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
9880 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
9882 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
9883 terminated STRING_CST object.
9885 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
9887 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
9889 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
9891 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
9892 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
9893 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
9894 the elements into a register.
9895 (rs6000_split_v4si_init_di_reg): Delete.
9896 (rs6000_split_v4si_init): Delete.
9897 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
9898 (vsx_init_v4si): Rewrite as a define_expand.
9900 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
9902 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
9903 zero_extend argument from memory): New.
9905 2018-07-22 Martin Sebor <msebor@redhat.com>
9908 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
9909 diagnosing calls with unknown arguments unless -Walloca-larger-than
9910 is restricted to less than PTRDIFF_MAX bytes.
9912 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
9914 * doc/gcov.texi (Invoking Gcov): Editorial changes.
9916 2018-07-20 David Malcolm <dmalcolm@redhat.com>
9918 * pretty-print.c (text_info::set_location): Remove redundant
9919 "line_table" parameter from call to rich_location::set_range.
9921 2018-07-20 Martin Sebor <msebor@redhat.com>
9924 * builtins.c (expand_builtin_alloca): Adjust.
9925 * calls.c (alloc_max_size): Simplify.
9926 * cgraphunit.c (cgraph_node::expand): Adjust.
9927 * common.opt (larger_than_size, warn_frame_larger_than): Remove
9929 (frame_larger_than_size): Same.
9930 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
9931 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
9933 * doc/invoke.texi (GCC Command Options): Document option arguments.
9934 Explain byte-size arguments and suffixes.
9935 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
9936 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
9937 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
9938 * doc/options.texi (UInteger): Expand.
9939 (Host_Wide_Int, ByteSize): Document new properties.
9940 * final.c (final_start_function_1): Include sizes in an error message.
9941 * function.c (frame_offset_overflow): Same.
9942 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
9943 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
9944 Diagnose unbounded alloca calls only for limits of less than
9946 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
9947 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
9949 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
9950 only for limits of less than PTRDIFF_MAX.
9951 * langhooks-def.h (lhd_handle_option): Change function argument
9953 * langhooks.c (lhd_handle_option): Same.
9954 * langhooks.h (handle_option): Same.
9955 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
9957 (var_type, var_type_struct): Same.
9958 (var_set): Handle ByteSize flag.
9959 * optc-gen.awk: Add comments to output to ease debugging. Make
9960 use of HOST_WIDE_INT where appropriate.
9961 * opts-gen-save.awk: Use %lx to format unsigned long.
9962 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
9963 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
9964 arguments. Parse bytes-size suffixes.
9965 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
9966 (enum_value_to_arg): Same.
9967 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
9968 (handle_option): Adjust.
9969 (generate_option): Change function argument to HOST_WIDE_INT.
9970 (cmdline_handle_error): Adjust.
9971 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
9972 (set_option): Change function argument to HOST_WIDE_INT.
9973 (option_enabled): Handle cl_host_wide_int.
9974 (get_option_state): Handle CLVC_SIZE.
9975 (control_warning_option): Same.
9976 * opts.c (common_handle_option): Change function argument to
9977 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
9978 OPT_Wvla_larger_than_.
9979 * opts.h (enum cl_var_type): Add an enumerator.
9980 * stor-layout.c (layout_decl): Print a more meaningful warning.
9981 * toplev.c (output_stack_usage): Adjust.
9983 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
9985 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
9986 call to inline_expand_builtin_string_cmp.
9987 (expand_builtin_strcmp): Likewise.
9988 (expand_builtin_strncmp): Likewise.
9989 (inline_string_cmp): Delete the last parameter, change char_type_node
9990 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
9992 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
9993 the inlining expansion on target where the type of the call has same or
9994 narrower precision than unsigned char.
9996 2018-07-20 David Malcolm <dmalcolm@redhat.com>
9998 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
9999 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
10000 * common.opt (fsave-optimization-record): New option.
10001 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
10002 * doc/invoke.texi (-fsave-optimization-record): New option.
10003 * dumpfile.c: Include "optinfo-emit-json.h".
10004 (struct kv_pair): Move to coretypes.h.
10005 (optgroup_options): Make non-static.
10006 (dump_context::end_scope): Call
10007 optimization_records_maybe_pop_dump_scope.
10008 * dumpfile.h (optgroup_options): New decl.
10009 * json.cc: New file.
10010 * json.h: New file.
10011 * optinfo-emit-json.cc: New file.
10012 * optinfo-emit-json.h: New file.
10013 * optinfo.cc: Include "optinfo-emit-json.h".
10014 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
10015 (optinfo_enabled_p): Check optimization_records_enabled_p.
10016 (optinfo_wants_inlining_info_p): Likewise.
10017 * optinfo.h: Update comment.
10018 * profile-count.c (profile_quality_as_string): New function.
10019 * profile-count.h (profile_quality_as_string): New decl.
10020 (profile_count::quality): New accessor.
10021 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
10022 and optinfo_emit_json_cc_tests.
10023 * selftest.h (selftest::json_cc_tests): New decl.
10024 (selftest::optinfo_emit_json_cc_tests): New decl.
10025 * toplev.c: Include "optinfo-emit-json.h".
10026 (compile_file): Call optimization_records_finish.
10027 (do_compile): Call optimization_records_start.
10028 * tree-ssa-live.c: Include optinfo.h.
10029 (remove_unused_scope_block_p): Retain inlining information if
10030 optinfo_wants_inlining_info_p returns true.
10032 2018-07-20 Richard Biener <rguenther@suse.de>
10035 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
10036 to cover -flto-partition=none.
10038 2018-07-20 Martin Liska <mliska@suse.cz>
10040 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
10041 (get_decl_source_range): Remove unused function.
10043 2018-07-20 Richard Biener <rguenther@suse.de>
10045 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
10046 (struct vn_phi_s): Likewise.
10047 (struct vn_reference_s): Likewise.
10048 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
10049 for searching the slot of an entry known to be in the hash itself.
10050 (vn_phi_hasher::equal): Likewise.
10051 (vn_reference_hasher::equal): Likewise.
10052 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
10054 (optimistic_info, current_info): Remove, keeping only valid_info.
10055 (vn_reference_lookup_1): Remove fallback lookup.
10056 (vn_reference_lookup_2): Likewise.
10057 (vn_nary_op_lookup_1): Likewise.
10058 (vn_phi_lookup): Likewise.
10059 (vn_nary_build_or_lookup_1): Make sure to not chain the built
10061 (vn_reference_insert): Adjust, chain the inserted hash element
10062 at last_inserted_ref.
10063 (vn_reference_insert_pieces): Likewise.
10064 (visit_reference_op_call): Likewise.
10065 (vn_nary_op_insert_into): Chain the inserted hash element at
10066 last_inserted_nary.
10067 (vn_nary_op_insert_pieces): Adjust.
10068 (vn_nary_op_insert): Likewise.
10069 (vn_nary_op_insert_stmt): Likewise.
10070 (vn_phi_insert): Adjust, chain the inserted hash element at
10072 (process_scc): Remove clearing and copying the optimistic
10073 table. Instead remove elements inserted during an optimistic
10074 iteration from the single table we maintain.
10075 (init_scc_vn): Adjust.
10076 (free_scc_vn): Likewise.
10077 (sccvn_dom_walker::record_cond): Likewise.
10078 (sccvn_dom_walker::after_dom_children): Likewise.
10080 2018-07-19 Martin Sebor <msebor@redhat.com>
10082 PR tree-optimization/84047
10083 PR tree-optimization/83776
10084 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
10085 (check_array_bounds): Call it.
10087 2018-07-19 Martin Sebor <msebor@redhat.com>
10089 * align.h (align_flags): Use member initialization.
10091 2018-07-19 David Malcolm <dmalcolm@redhat.com>
10093 * Makefile.in (OBJS): Add optinfo.o.
10094 * coretypes.h (class symtab_node): New forward decl.
10095 (struct cgraph_node): New forward decl.
10096 (class varpool_node): New forward decl.
10097 * dump-context.h: New file.
10098 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
10100 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
10101 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
10102 (set_alt_dump_file): Likewise.
10103 (dump_context::~dump_context): New dtor.
10104 (dump_gimple_stmt): Move implementation to...
10105 (dump_context::dump_gimple_stmt): ...this new member function.
10106 Add the stmt to any pending optinfo, creating one if need be.
10107 (dump_gimple_stmt_loc): Move implementation to...
10108 (dump_context::dump_gimple_stmt_loc): ...this new member function.
10109 Start a new optinfo and add the stmt to it.
10110 (dump_gimple_expr): Move implementation to...
10111 (dump_context::dump_gimple_expr): ...this new member function.
10112 Add the stmt to any pending optinfo, creating one if need be.
10113 (dump_gimple_expr_loc): Move implementation to...
10114 (dump_context::dump_gimple_expr_loc): ...this new member function.
10115 Start a new optinfo and add the stmt to it.
10116 (dump_generic_expr): Move implementation to...
10117 (dump_context::dump_generic_expr): ...this new member function.
10118 Add the tree to any pending optinfo, creating one if need be.
10119 (dump_generic_expr_loc): Move implementation to...
10120 (dump_context::dump_generic_expr_loc): ...this new member
10121 function. Add the tree to any pending optinfo, creating one if
10123 (dump_printf): Move implementation to...
10124 (dump_context::dump_printf_va): ...this new member function. Add
10125 the text to any pending optinfo, creating one if need be.
10126 (dump_printf_loc): Move implementation to...
10127 (dump_context::dump_printf_loc_va): ...this new member function.
10128 Start a new optinfo and add the stmt to it.
10129 (dump_dec): Move implementation to...
10130 (dump_context::dump_dec): ...this new member function. Add the
10131 value to any pending optinfo, creating one if need be.
10132 (dump_context::dump_symtab_node): New member function.
10133 (dump_context::get_scope_depth): New member function.
10134 (dump_context::begin_scope): New member function.
10135 (dump_context::end_scope): New member function.
10136 (dump_context::ensure_pending_optinfo): New member function.
10137 (dump_context::begin_next_optinfo): New member function.
10138 (dump_context::end_any_optinfo): New member function.
10139 (dump_context::s_current): New global.
10140 (dump_context::s_default): New global.
10141 (dump_scope_depth): Delete global.
10142 (dumpfile_ensure_any_optinfo_are_flushed): New function.
10143 (dump_symtab_node): New function.
10144 (get_dump_scope_depth): Reimplement in terms of dump_context.
10145 (dump_begin_scope): Likewise.
10146 (dump_end_scope): Likewise.
10147 (selftest::temp_dump_context::temp_dump_context): New ctor.
10148 (selftest::temp_dump_context::~temp_dump_context): New dtor.
10149 (selftest::verify_item): New function.
10150 (ASSERT_IS_TEXT): New macro.
10151 (ASSERT_IS_TREE): New macro.
10152 (ASSERT_IS_GIMPLE): New macro.
10153 (selftest::test_capture_of_dump_calls): New test.
10154 (selftest::dumpfile_c_tests): Call it.
10155 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
10156 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
10157 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
10158 descriptive comment.
10159 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
10160 (dump_node, dump_bb): Move these unrelated decls.
10161 (class dump_manager): Add leading comment.
10162 * optinfo.cc: New file.
10163 * optinfo.h: New file.
10165 2018-07-19 Michael Collison <michael.collison@arm.com>
10166 Richard Henderson <rth@redhat.com>
10168 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
10169 (subti): Handle op1 zero.
10170 (subvti4, usub4ti4): New.
10171 (*sub<GPI>3_compare1_imm): New.
10172 (sub<GPI>3_carryinCV): New.
10173 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
10174 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
10176 2018-07-19 Michael Collison <michael.collison@arm.com>
10177 Richard Henderson <rth@redhat.com>
10179 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
10180 (addti3): Create simpler code if low part is already known to be 0.
10181 (addvti4, uaddvti4): New.
10182 (*add<GPI>3_compareC_cconly_imm): New.
10183 (*add<GPI>3_compareC_cconly): New.
10184 (*add<GPI>3_compareC_imm): New.
10185 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
10186 handle constants within this pattern..
10187 (*add<GPI>3_compareV_cconly_imm): New.
10188 (*add<GPI>3_compareV_cconly): New.
10189 (*add<GPI>3_compareV_imm): New.
10190 (add<GPI>3_compareV): New.
10191 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
10192 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
10193 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
10194 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
10195 with 'comparison' operator.
10196 (*add<GPI>3_compareV_cconly_imm): Ditto.
10197 (*add<GPI>3_compareV_cconly): Ditto.
10198 (*add<GPI>3_compareV_imm): Ditto.
10199 (add<GPI>3_compareV): Ditto.
10200 (add<mode>3_carryinC): Ditto.
10201 (*add<mode>3_carryinC_zero): Ditto.
10202 (*add<mode>3_carryinC): Ditto.
10203 (add<mode>3_carryinV): Ditto.
10204 (*add<mode>3_carryinV_zero): Ditto.
10205 (*add<mode>3_carryinV): Ditto.
10207 2018-07-19 Michael Collison <michael.collison@arm.com>
10208 Richard Henderson <rth@redhat.com>
10210 * config/aarch64/aarch64-modes.def (CC_V): New.
10211 * config/aarch64/aarch64-protos.h
10212 (aarch64_addti_scratch_regs): Declare
10213 (aarch64_subvti_scratch_regs): Declare.
10214 (aarch64_expand_subvti): Declare.
10215 (aarch64_gen_unlikely_cbranch): Declare
10216 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
10217 for signed overflow using CC_Vmode.
10218 (aarch64_get_condition_code_1): Handle CC_Vmode.
10219 (aarch64_gen_unlikely_cbranch): New function.
10220 (aarch64_addti_scratch_regs): New function.
10221 (aarch64_subvti_scratch_regs): New function.
10222 (aarch64_expand_subvti): New function.
10224 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
10226 * config/aarch64/aarch64-option-extensions.def: New entry for profile
10228 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
10229 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
10232 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
10235 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
10236 address check not strict.
10238 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
10240 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
10242 (store_pair_lanes<mode>): Likewise.
10243 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
10244 enum value 'ADDR_QUERY_LDP_STP_N'.
10245 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
10246 (aarch64_print_address_internal): Add declaration.
10247 (aarch64_print_ldpstp_address): Remove.
10248 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
10249 (aarch64_print_operand): Change printing of 'y'.
10250 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
10251 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
10252 'true' rather than '1'.
10253 * config/aarch64/constraints.md (Uml): Likewise.
10254 (Uml): Rename to Umn.
10257 2018-07-19 Richard Biener <rguenther@suse.de>
10259 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
10261 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
10262 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
10263 (vn_reference_hasher): Likewise.
10264 (struct vn_tables_s): Remove obstack and alloc-pool members.
10265 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
10266 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
10267 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
10268 (vn_reference_insert_pieces): Likewise.
10269 (alloc_vn_nary_op_noinit): Adjust.
10270 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
10271 (vn_phi_eq): Adjust.
10272 (shared_lookup_phiargs): Remove.
10273 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
10274 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
10275 (visit_reference_op_call): Likewise.
10276 (copy_nary, copy_phi, copy_reference): Remove.
10277 (process_scc): Rewind the obstack when iterating. Do not
10278 copy the elements to valid_info but just move them from one
10279 hashtable to the other.
10280 (allocate_vn_table): Adjust.
10281 (free_vn_table): Likewise.
10282 (init_scc_vn): Likewise.
10283 (free_scc_vn): Likewise.
10285 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
10288 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
10289 indirect_return as function type attribute.
10290 (ix86_attribute_table): Change indirect_return to function
10292 * doc/extend.texi: Update indirect_return attribute.
10294 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
10296 * wide-int.h (widest2_int): New.
10297 * gimple-fold.c (arith_overflowed_p): Use it.
10298 * tree.h (widest2_int_cst): New.
10299 * tree-vrp.c (wide_int_binop_overflow): Rename from
10300 vrp_int_const_binop.
10301 Rewrite to work on trees.
10302 (extract_range_from_multiplicative_op_1): Abstract code to...
10303 (wide_int_range_min_max): ...here.
10304 (wide_int_range_cross_product): ...and here.
10305 (extract_range_from_binary_expr_1): Abstract overflow code to...
10306 (wide_int_range_mult_wrapping): ...here.
10307 * tree-vrp.h (wide_int_range_cross_product): New.
10308 (wide_int_range_mult_wrapping): New.
10310 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
10311 Julia Koval <julia.koval@intel.com>
10313 * config/i386/x86-tune-costs.h (skylake_memcpy,
10314 skylake_memset): Replace rep_prefix with unrolling for size 512.
10316 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
10318 PR middle-end/86544
10319 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
10320 comparision with EQ_EXPR in last stmt.
10322 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
10324 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
10325 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
10326 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
10327 previously known as "PowerPC AltiVec Built-in Functions". Move
10328 some material to new subsubsections "PowerPC AltiVec Built-in
10329 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
10331 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
10332 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
10333 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
10334 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
10336 2018-07-18 Richard Biener <rguenther@suse.de>
10338 PR tree-optimization/86557
10339 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
10342 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
10344 * config/s390/s390.c (s390_function_profiler): Generate CFI.
10346 2018-07-17 Jeff Law <law@redhat.com>
10348 * config/arm/arm.c (get_label_padding): Update for recent
10349 changes to label_to_alignment.
10351 PR tree-optimization/86010
10352 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
10354 * config/mips/mips.c (vr4130_align_insns): Update for recent
10355 changes to label_to_alignment.
10357 * config/frv/frv.c (frv_label_align): Update for recent changes
10358 to label_to_alignment.
10360 * config/nios2/nios2.c (nios2_label_align): Update for recent
10361 changes which dropped ALIGN_LABELS_LOG.
10363 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
10365 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
10366 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
10368 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10370 * config/arc/arc.c (arc_label_align): Use align_labels instead of
10371 deprecated align_labels_log.
10373 2018-07-17 Richard Biener <rguenther@suse.de>
10376 * dwarf2out.c (init_sections_and_labels): Always generate
10377 a debug_line_str_section for early LTO debug.
10378 (dwarf2out_finish): Reset debug_line_str_hash output early.
10379 Bump counter for extra dwarf5 .debug_loc labels to not conflict
10381 (dwarf2out_early_finish): Output debug_line_str.
10383 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
10385 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
10386 index register on z196 or later.
10388 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
10390 * config/s390/s390.c (s390_default_align): Set default function
10392 (s390_override_options_after_change): Call s390_default align.
10393 (s390_option_override_internal): Call s390_default_align.
10394 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
10396 2018-07-17 Jakub Jelinek <jakub@redhat.com>
10398 PR middle-end/86542
10399 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
10400 to _looptemp_ clauses, other than the first two.
10402 2018-07-17 Martin Liska <mliska@suse.cz>
10404 * opts.c: Do not enable OPT_falign_* for -Os.
10406 2018-07-17 Martin Liska <mliska@suse.cz>
10408 * align.h (MAX_CODE_ALIGN): New.
10409 (MAX_CODE_ALIGN_VALUE): New.
10410 * common/config/i386/i386-common.c (ix86_handle_option):
10411 (MAX_CODE_ALIGN): Moved to align.h.
10412 * final.c (MAX_CODE_ALIGN): Likewise.
10413 * opts.c (parse_and_check_align_values):
10414 (MAX_CODE_ALIGN): Likewise.
10415 (MAX_CODE_ALIGN_VALUE): Likewise.
10417 2018-07-17 Martin Liska <mliska@suse.cz>
10419 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
10420 in order to fulfil coding style.
10421 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
10422 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10423 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
10424 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
10425 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
10426 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
10427 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
10428 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
10429 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10431 2018-07-17 Martin Liska <mliska@suse.cz>
10433 * align.h: New file.
10434 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
10436 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
10437 align_flags of label_to_alignment.
10438 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
10440 * config/m68k/m68k.c: Do not use removed align_labels_value and
10442 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
10443 (LOOP_ALIGN): Likewise.
10444 (LABEL_ALIGN): Likewise.
10445 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
10446 Remove not used macro.
10447 (rs6000_loop_align): Change return type to align_flags.
10448 (rs6000_loop_align_max_skip): Remove.
10449 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
10450 Change return type to align_flags.
10451 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
10452 Remove not used macro.
10453 (rs6000_loop_align): Change return type to align_flags.
10454 (rs6000_loop_align_max_skip): Remove.
10455 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
10456 * config/rx/rx-protos.h (rx_align_for_label): Make it
10458 * config/rx/rx.c (rx_align_for_label): Change return type
10460 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
10462 into align_flags class.
10463 (LABEL_ALIGN): Likewise.
10464 (LOOP_ALIGN): Likewise.
10465 * config/s390/s390.c (s390_label_align): Use align_flags
10467 (s390_asm_output_function_label): Likewise.
10468 * config/sh/sh.c (sh_override_options_after_change):
10469 Use align_flags class directly without macros.
10470 (find_barrier): Likewise.
10471 (barrier_align): Likewise.
10472 (sh_loop_align): Likewise.
10473 * config/spu/spu.c (spu_option_override):
10474 Use align_flags_tuple::get_value instead of removed macros.
10475 (spu_sched_init): Likewise.
10476 * config/spu/spu.h (GTY): Likewise.
10477 * config/visium/visium.c (visium_option_override):
10478 Set "8" as default secondary alignment.
10479 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
10480 in order to guarantee secondary alignment of 8.
10481 * coretypes.h: Include align.h header file.
10482 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
10483 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
10484 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
10485 * doc/tm.texi.in: Likewise.
10486 * final.c (struct label_alignment): Remove not used structure.
10487 (LABEL_ALIGN): Change type to align_flags.
10488 (LOOP_ALIGN): Likewise.
10489 (JUMP_ALIGN): Likewise.
10490 (default_loop_align_max_skip): Remove.
10491 (default_label_align_max_skip): Likewise.
10492 (default_jump_align_max_skip): Likewise.
10493 (default_label_align_after_barrier_max_skip):
10494 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
10495 (LABEL_TO_MAX_SKIP): Remove.
10496 (label_to_alignment): Return align_flags type instead of integer.
10497 (label_to_max_skip): Remove.
10498 (align_fuzz): Use align_flags type.
10499 (compute_alignments): Use align_flags type and use align_flags::max
10500 to combine multiple alignments.
10501 (grow_label_align): Grow vec instead of C array.
10502 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
10503 (shorten_branches): Use align_flags type and use align_flags::max
10504 to combine multiple alignments.
10505 (final_scan_insn_1): Remove usage of secondary alignment that comes
10506 from label alignment, but instead use proper secondary alignment
10507 which is computed in grow_label_align.
10508 * flags.h (struct align_flags_tuple): Move to align.h.
10509 (struct align_flags): Likewise.
10510 (state_align_loops): Rename to align_loops.
10511 (state_align_jumps): Rename to align_jumps.
10512 (state_align_labels): Rename to align_labels.
10513 (state_align_functions): Rename to align_functions.
10514 (align_loops_log): Remove.
10515 (align_jumps_log): Remove.
10516 (align_labels_log): Remove.
10517 (align_functions_log): Remove.
10518 (align_loops_max_skip): Remove.
10519 (align_jumps_max_skip): Remove.
10520 (align_labels_max_skip): Remove.
10521 (align_functions_max_skip): Remove.
10522 (align_loops_value): Remove.
10523 (align_jumps_value): Remove.
10524 (align_labels_value): Remove.
10525 (align_functions_value): Remove.
10526 * output.h (label_to_alignment): Change return type to align_flags.
10527 (label_to_max_skip): Remove.
10528 * target.def: Remove loop_align_max_skip, label_align_max_skip,
10529 jump_align_max_skip macros.
10530 * targhooks.h (default_loop_align_max_skip): Remove.
10531 (default_label_align_max_skip): Likewise.
10532 (default_jump_align_max_skip): Likewise.
10533 (default_label_align_after_barrier_max_skip): Remove.
10534 * toplev.c (read_log_maxskip): Use ::normalize function.
10535 (parse_N_M): Remove not used argument and also call ::normalize.
10536 (parse_alignment_opts): Do not pass unused arguments.
10537 * varasm.c (assemble_start_function): Use directly align_functions
10538 instead of removed macros.
10539 * system.h: Do not poison removed macros.
10541 2018-07-17 Jakub Jelinek <jakub@redhat.com>
10543 PR middle-end/86539
10544 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
10545 and cond temporaries don't have reference type if iterator has
10546 pointer type. For init use &for_pre_body instead of pre_p if
10547 for_pre_body is non-empty.
10549 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
10551 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
10552 double-double modes to SFmode directly directly.
10553 (trunc<mode>sf2_fprs): Delete.
10555 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
10557 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
10558 for conversions between IFmode and the decimal floating point modes.
10559 (init_float128_ieee): Use the correct names for conversions between
10560 KFmode and the decimal floating point modes.
10562 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
10564 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
10565 for the conversions between TDmode and IFmode.
10566 (init_float128_ieee): Use more correct names for the conversions
10567 between TDmode and KFmode.
10569 2018-07-16 Jakub Jelinek <jakub@redhat.com>
10571 PR tree-optimization/86526
10572 * builtins.c (expand_builtin_memcmp): Formatting fixes.
10573 (inline_expand_builtin_string_cmp): Likewise.
10574 (inline_string_cmp): Likewise. Use c_readstr instead of
10575 builtin_memcpy_read_str. Add unit_mode temporary.
10577 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
10579 PR middle-end/86528
10580 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
10581 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
10583 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
10585 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
10586 Alphabetize prototypes of built-in functions, separating out
10587 built-in functions that are listed in this section but should be
10588 described elsewhere.
10590 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
10593 * expmed.c (emit_store_flag): Do not emit setcc followed by a
10594 conditional move when trapping comparison was split to a
10595 non-trapping one (and vice versa).
10597 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
10599 * config/s390/s390.c (s390_function_profiler): Generate nops
10600 instead of profiler call sequences.
10601 * config/s390/s390.opt: Add the new option.
10603 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
10605 * config/s390/s390.c (s390_function_profiler): Generate
10606 __mcount_loc section.
10607 * config/s390/s390.opt: Add the new option.
10609 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
10611 * common.opt: Add the new warning.
10612 * config/s390/s390.c (s390_function_profiler): Emit "brasl
10613 %r0,__fentry__" when -mfentry is specified.
10614 (s390_option_override_internal): Disallow -mfentry for 31-bit
10616 * config/s390/s390.opt: Add the new option.
10618 2018-07-16 Richard Biener <rguenther@suse.de>
10621 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
10622 for function-local FUNCTION_DECL and RESULT_DECL immediately.
10624 2018-07-16 Martin Liska <mliska@suse.cz>
10627 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
10630 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
10632 * config/arc/arcHS.md: Update ARCHS scheduling rules.
10634 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
10636 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
10638 * config/arc/arc-cpus.def (hs4x): New cpu.
10640 * config/arc/arc-tables.opt: Regenerate.
10641 * config/arc/arc.c (arc_sched_issue_rate): New function.
10642 (TARGET_SCHED_ISSUE_RATE): Define.
10643 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
10644 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
10646 (attr tune): Add ARCHS4x tune values.
10647 (attr tune_dspmpy): Define.
10648 (*tst): Correct instruction type.
10649 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
10650 * config/arc/arcHS4x.md: New file.
10651 * config/arc/fpu.md: Update instruction type attributes.
10652 * config/arc/t-multilib: Regenerate.
10654 2018-07-16 Tom de Vries <tdevries@suse.de>
10657 * var-tracking.c (vt_initialize): Fix pre_dec handling.
10659 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10661 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
10664 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
10666 PR tree-optimization/86514
10667 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
10668 conversion to a boolean type from a type with greater precision.
10670 2018-07-16 Tom de Vries <tdevries@suse.de>
10672 * var-tracking.c (vt_initialize): Print adjusted insn slim if
10673 dump_flags request TDF_SLIM.
10675 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
10677 * fold-const.c (int_const_binop_1): Abstract...
10678 (wide_int_binop): ...wide int code here.
10679 (poly_int_binop): ...poly int code here.
10680 Abstract the rest of int_const_binop_1 into int_const_binop.
10681 * fold-const.h (wide_int_binop): New.
10682 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
10683 Remove useless PLUS/MINUS_EXPR case.
10684 (zero_nonzero_bits_from_vr): Move wide int code...
10685 (zero_nonzero_bits_from_bounds): ...here.
10686 (extract_range_from_binary_expr_1): Move mask optimization code...
10687 (range_easy_mask_min_max): ...here.
10688 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
10689 (range_easy_mask_min_max): New.
10691 2018-07-15 Jeff Law <law@redhat.com>
10694 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
10697 2018-07-14 Jim Wilson <jimw@sifive.com>
10699 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
10701 2018-07-14 Paul Koning <ni1d@arrl.net>
10703 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
10705 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
10707 * lto-streamer-out.c (copy_function_or_variable): Dump info about
10710 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
10711 Steve Munroe <munroesj52@gmail.com>
10713 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
10714 (_mm_andnot_si128): Likewise.
10715 (_mm_or_si128): Likewise.
10716 (_mm_xor_si128): Likewise.
10718 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
10720 PR middle-end/78809
10721 * builtins.c (expand_builtin_memcmp): Inline the calls first
10722 when result_eq is false.
10723 (expand_builtin_strcmp): Inline the calls first.
10724 (expand_builtin_strncmp): Likewise.
10725 (inline_string_cmp): New routine. Expand a string compare
10726 call by using a sequence of char comparison.
10727 (inline_expand_builtin_string_cmp): New routine. Inline expansion
10728 a call to str(n)cmp/memcmp.
10729 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
10731 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
10733 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
10735 * config/arm/driver-arm.c: Include arm-native.h.
10736 (host_detect_local_cpu): Use auto-generated data tables.
10737 (vendors, arm_cpu_table): Delete. Move part information to ...
10738 * config/arm/arm-cpus.in: ... here.
10739 * config/arm/parsecpu.awk (gen_native): New function.
10740 (vendor, part): New CPU fields.
10741 (END): Add support for building the native CPU detection tables.
10742 * config/arm/t-arm (arm-native.h): Add build rule.
10743 (driver-arm.o): Add dependency on arm-native.h.
10745 2018-07-13 Richard Biener <rguenther@suse.de>
10747 PR middle-end/85974
10748 * match.pd (addr1 - addr2): Allow either of the operand to
10751 2018-07-13 Tom de Vries <tdevries@suse.de>
10753 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
10756 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
10758 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
10759 arrays instead of numbered variables.
10761 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
10763 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
10764 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
10766 2018-07-13 Richard Biener <rguenther@suse.de>
10769 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
10770 instead of get_context_die.
10772 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
10773 Richard Biener <rguenther@suse.de>
10775 PR middle-end/86489
10776 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
10777 that the loop latch destination where phi is defined.
10779 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
10781 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
10782 (riscv_expand_epilogue): Add assertion to check interrupt mode.
10783 (riscv_set_current_function): Extract getting interrupt type to new
10785 (riscv_get_interrupt_type): New function.
10786 (riscv_merge_decl_attributes): New function, checking interrupt type is
10788 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
10790 2018-07-12 Paul Koning <ni1d@arrl.net>
10792 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
10795 2018-07-12 Paul Koning <ni1d@arrl.net>
10797 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
10798 zero reference, add doloop_end instead.
10799 * doc/md.texi (decrement_and_branch_until_zero): Remove.
10800 (Looping patterns): Remove decrement_and_branch_until_zero. Add
10801 detail for doloop_end.
10803 2018-07-12 Martin Sebor <msebor@redhat.com>
10806 * attribs.c (decl_attributes): Reject conflicting attributes before
10807 calling attribute handlers.
10809 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
10811 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
10813 (gcc::dump_manager::get_dump_file_name): likewise.
10814 (dump_begin): Likewise.
10815 * dumpfile.h (dump_begin): Update prototype.
10816 (gcc::dump_manager::get_dump_file_name,
10817 gcc::dump_manager::get_dump_file_name): Update prototype.
10819 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10821 * internal-fn.h (vectorizable_internal_fn_p): New function.
10822 * tree-vect-slp.c (compatible_calls_p): Likewise.
10823 (vect_build_slp_tree_1): Remove nops argument. Handle calls
10824 to internal functions.
10825 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
10827 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10829 * fold-const.h (inverse_conditions_p): Declare.
10830 * fold-const.c (inverse_conditions_p): New function.
10831 * match.pd: Use inverse_conditions_p. Add folds of view_converts
10832 that test the inverse condition of a conditional internal function.
10833 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
10834 * internal-fn.c (internal_fn_mask_index): Handle conditional
10835 internal functions.
10836 (vectorized_internal_fn_supported_p): New function.
10837 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
10838 (any_pred_load_store): Replace with...
10839 (need_to_predicate): ...this new variable.
10840 (redundant_ssa_names): New variable.
10841 (ifcvt_can_use_mask_load_store): Move initial checks to...
10842 (ifcvt_can_predicate): ...this new function. Handle tree codes
10843 for which a conditional internal function exists.
10844 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
10845 instead of ifcvt_can_use_mask_load_store. Update after variable
10847 (predicate_load_or_store): New function, split out from
10848 predicate_mem_writes.
10849 (check_redundant_cond_expr): New function.
10850 (value_available_p): Likewise.
10851 (predicate_rhs_code): Likewise.
10852 (predicate_mem_writes): Rename to...
10853 (predicate_statements): ...this. Use predicate_load_or_store
10854 and predicate_rhs_code.
10855 (combine_blocks, tree_if_conversion): Update after above name changes.
10856 (ifcvt_local_dce): Handle redundant_ssa_names.
10857 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
10858 general conditional functions.
10859 * tree-vect-stmts.c (vectorizable_call): Likewise.
10861 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10862 Alan Hayward <alan.hayward@arm.com>
10863 David Sherwood <david.sherwood@arm.com>
10865 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
10866 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
10867 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
10868 plus and minus and convert them into IFN_COND_FMA-based sequences.
10869 (convert_mult_to_fma): Handle conditional plus and minus.
10871 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10873 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
10874 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
10875 (cond_fnms_optab): New optabs.
10876 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
10877 internal functions.
10878 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
10879 * internal-fn.h (get_conditional_internal_fn): Declare.
10880 (get_unconditional_internal_fn): Likewise.
10881 * internal-fn.c (cond_ternary_direct): New macro.
10882 (expand_cond_ternary_optab_fn): Likewise.
10883 (direct_cond_ternary_optab_supported_p): Likewise.
10884 (FOR_EACH_COND_FN_PAIR): Likewise.
10885 (get_conditional_internal_fn): New function.
10886 (get_unconditional_internal_fn): Likewise.
10887 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
10888 (gimple_match_op::gimple_match_op): Add a new overload for 5
10890 (gimple_match_op::set_op): Likewise.
10891 (gimple_resimplify5): Declare.
10892 * genmatch.c (decision_tree::gen): Generate simplifications for
10894 * gimple-match-head.c (gimple_simplify): Define an overload for
10895 5 operands. Handle calls with 5 arguments in the top-level overload.
10896 (convert_conditional_op): Handle conversions from unconditional
10897 internal functions to conditional ones.
10898 (gimple_resimplify5): New function.
10899 (build_call_internal): Pass a fifth operand.
10900 (maybe_push_res_to_seq): Likewise.
10901 (try_conditional_simplification): Try converting conditional
10902 internal functions to unconditional internal functions.
10903 Handle 3-operand unconditional forms.
10904 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
10905 Define ternary equivalents of the current rules for binary conditional
10906 internal functions.
10907 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
10908 ternary operations.
10909 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
10910 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
10911 (optab): Handle them.
10912 (SVE_COND_FP_TERNARY): New int iterator.
10913 (sve_fmla_op, sve_fmad_op): New int attributes.
10914 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
10915 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
10916 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
10918 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10920 * target.def (preferred_else_value): New target hook.
10921 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
10922 * doc/tm.texi: Regenerate.
10923 * targhooks.h (default_preferred_else_value): Declare.
10924 * targhooks.c (default_preferred_else_value): New function.
10925 * internal-fn.h (conditional_internal_fn_code): Declare.
10926 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
10927 (get_conditional_internal_fn): Use it.
10928 (conditional_internal_fn_code): New function.
10929 * gimple-match.h (gimple_match_cond): New struct.
10930 (gimple_match_op): Add a cond member function.
10931 (gimple_match_op::gimple_match_op): Update all forms to take a
10933 * genmatch.c (expr::gen_transform): Use the same condition as res_op
10934 for the suboperation, but don't specify a particular else_value.
10935 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
10936 (visit_nary_op, visit_reference_op_load): Pass
10937 gimple_match_cond::UNCOND to the gimple_match_op constructor.
10938 * gimple-match-head.c: Include tree-eh.h
10939 (convert_conditional_op): New function.
10940 (maybe_resimplify_conditional_op): Likewise.
10941 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
10942 (gimple_resimplify2): Likewise.
10943 (gimple_resimplify3): Likewise.
10944 (gimple_resimplify4): Likewise.
10945 (maybe_push_res_to_seq): Return null for conditional operations.
10946 (try_conditional_simplification): New function.
10947 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
10949 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
10951 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
10953 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
10955 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
10957 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
10959 (hash_tree): Do not hash DECL_FCONTEXT
10960 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
10961 Do not stream DECL_FCONTEXT.
10962 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
10963 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
10965 2018-07-12 Richard Biener <rguenther@suse.de>
10968 * dwarf2out.c (gen_block_die): Only output blocks when they have
10969 at least one !DECL_IGNORED_P variable.
10971 2018-07-12 Richard Biener <rguenther@suse.de>
10974 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
10975 Remove -mieee-fp handling.
10977 2018-07-12 Richard Biener <rguenther@suse.de>
10979 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
10980 left-over from last patch.
10982 2018-07-12 Jakub Jelinek <jakub@redhat.com>
10984 PR tree-optimization/86492
10985 * gimple-ssa-store-merging.c
10986 (imm_store_chain_info::coalesce_immediate_stores): Call
10987 check_no_overlap even for the merge_overlapping case. Formatting fix.
10989 2018-07-12 Richard Biener <rguenther@suse.de>
10991 PR middle-end/86479
10992 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
10993 move possibly trapping operations into the conditional.
10995 2018-07-12 Richard Biener <rguenther@suse.de>
10997 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
10998 (vn_lookup_simplify_result): Remove recursion limit applied
11000 (vn_nary_build_or_lookup_1): Adjust.
11001 (try_to_simplify): Likewise.
11002 * gimple-match-head.c (gimple_resimplify1): Instead apply one
11004 (gimple_resimplify2): Likewise.
11005 (gimple_resimplify3): Likewise.
11006 (gimple_resimplify4): Likewise.
11008 2018-07-11 Jakub Jelinek <jakub@redhat.com>
11010 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
11011 Use __mmask64 type instead of __mmask8 for __M argument.
11012 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
11013 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
11015 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
11016 __mmask16 for __M argument.
11017 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
11018 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
11019 to __mmask16 instead of __mmask8.
11020 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
11021 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
11022 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
11023 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
11024 instead of __mmask16 for __U argument.
11025 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
11026 __mmask16 instead of __mmask8 for __U argument.
11027 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
11029 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
11031 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
11033 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
11035 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
11037 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
11039 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
11041 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
11043 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
11045 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
11047 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
11048 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
11049 return type as well as __M argument type and all casts from __mmask8
11051 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
11052 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
11053 return type as well as __M argument type and all casts from __mmask8
11055 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
11056 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
11057 return type as well as __M argument type and all casts from __mmask8
11059 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
11060 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
11061 return type as well as __M argument type and all casts from __mmask8
11063 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
11064 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
11067 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
11069 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
11070 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
11073 2018-07-11 Paul Koning <ni1d@arrl.net>
11075 * doc/md.texi (define_subst): Document how multiple occurrences of
11076 the same argument in the replacement pattern are handled.
11078 2018-07-11 Paul Koning <ni1d@arrl.net>
11080 * doc/extend.texi (Common Variable Attributes): Move "mode" into
11081 alphabetical order.
11082 (Common Type Attributes): Add "mode" attribute.
11084 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
11086 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
11087 stream DECL_ORIGINAL_TYPE.
11088 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
11089 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
11090 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
11091 Do not walk original type.
11092 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
11094 (write_ts_decl_non_common_tree_pointers): Do not stream
11096 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
11097 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
11099 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
11101 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
11102 thread twice from the same starting edge.
11104 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
11106 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
11107 * gimple.c (gimple_call_nonnull_result_p): ...here...
11108 (gimple_call_nonnull_arg): ...and here.
11109 * gimple.h (gimple_call_nonnull_result_p): New.
11110 (gimple_call_nonnull_arg): New.
11112 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
11114 * config/arm/arm-cpus.in: Move information from fpu field of each
11115 cpu definition to the isa field.
11116 * config/arm/parsecpu.awk (fpu): Delete match rule.
11117 (gen_comm_data): Don't add bits from the CPU's FPU entry.
11119 2018-07-11 Richard Biener <rguenther@suse.de>
11122 * dwarf2out.c (init_sections_and_labels): Use
11123 output_asm_line_debug_info consistently.
11124 (dwarf2out_early_finish): Likewise.
11125 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
11128 2018-07-11 Richard Biener <rguenther@suse.de>
11130 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
11131 Rework father_bb setting in a way to avoid propagating constants
11132 multiple times on a loop body.
11134 2018-07-10 Mark Wielaard <mark@klomp.org>
11137 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
11139 2018-07-10 Richard Biener <rguenther@suse.de>
11141 * hash-map.h (hash_map::iterator::operator*): Return
11142 references to key and value.
11144 2018-07-10 Jakub Jelinek <jakub@redhat.com>
11147 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
11148 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
11149 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
11150 (gimplify_omp_for): For composite loops, move outer
11151 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
11152 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
11153 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
11154 TREE_LIST for both the original class iterator and the "last" helper
11155 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
11156 loop, remember has_decl_expr from outer composite loops for the
11157 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
11159 2018-07-09 Martin Sebor <msebor@redhat.com>
11161 PR middle-end/77357
11162 PR middle-end/86428
11163 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
11164 accessing implicitly initialized array elements.
11165 * expr.c (string_constant): Handle string initializers of
11166 character arrays within aggregates.
11167 * gimple-fold.c (fold_array_ctor_reference): Add argument.
11168 Store element offset. As a special case, handle zero size.
11169 (fold_nonarray_ctor_reference): Same.
11170 (fold_ctor_reference): Add argument. Store subobject offset.
11171 * gimple-fold.h (fold_ctor_reference): Add argument.
11173 2018-07-09 Paul Koning <ni1d@arrl.net>
11175 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
11176 (pdp11_insn_cost): New function.
11177 (pdp11_md_asm_adjust): New function.
11178 (TARGET_INVALID_WITHIN_DOLOOP): Define.
11179 (pdp11_rtx_costs): Update to match machine better.
11180 (output_addr_const_pdp11): Correct format mismatch warnings.
11181 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
11182 * config/pdp11/pdp11.md: General change to add base_cost and/or
11183 length attributes for use by new pdp11_insn_cost function.
11184 (MIN_BRANCH): Correct definition.
11186 (doloop_end): Use standard pattern name for looping pattern.
11187 (doloop_end_nocc): New.
11188 (movsf): Add another constraint alternative.
11189 (zero_extendqihi2): Add constraint alternatives for not in place
11191 (zero_extendhisi2): Remove.
11192 (shift patterns): Add CC handling variants.
11196 (define_peephole2): New peephole to recognize mov that sets CC for
11199 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11201 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
11202 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
11205 2018-07-09 Jakub Jelinek <jakub@redhat.com>
11208 * real.c (real_nextafter): Return true if result is denormal.
11210 2018-07-09 Martin Liska <mliska@suse.cz>
11212 * common.opt: Add back wrongly removed attribute.
11214 2018-07-09 Richard Biener <rguenther@suse.de>
11217 * dwarf2out.c (gen_block_die): For an early generated DIE
11218 always output high/low PC attributes.
11220 2018-07-09 Tom de Vries <tdevries@suse.de>
11222 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
11223 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
11224 onto VAR_DECL with abstract origin.
11226 2018-07-07 Jim Wilson <jimw@sifive.com>
11228 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
11230 2018-07-07 Tom de Vries <tdevries@suse.de>
11232 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
11234 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
11236 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
11238 (combine_bound): Use wide-int overflow calculation instead of
11240 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
11242 * fold-const.c (int_const_binop_2): Same.
11243 (extract_muldiv_1): Same.
11244 (fold_div_compare): Same.
11245 (fold_abs_const): Same.
11247 * poly-int.h (add): Same.
11251 * predict.c (predict_iv_comparison): Same.
11252 * profile-count.c (slow_safe_scale_64bit): Same.
11253 * simplify-rtx.c (simplify_const_binary_operation): Same.
11254 * tree-chrec.c (tree_fold_binomial): Same.
11255 * tree-data-ref.c (split_constant_offset_1): Same.
11256 * tree-if-conv.c (idx_within_array_bound): Same.
11257 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
11258 * tree-ssa-phiopt.c (minmax_replacement): Same.
11259 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
11260 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
11261 * vr-values.c (vr_values::adjust_range_with_scev): Same.
11262 * wide-int.cc (wi::add_large): Same.
11263 (wi::mul_internal): Same.
11264 (wi::sub_large): Same.
11265 (wi::divmod_internal): Same.
11266 * wide-int.h: Change overflow type to overflow_type for neg, add,
11267 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
11268 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
11269 mul_internal, divmod_internal.
11270 (overflow_type): New enum.
11271 (accumulate_overflow): New.
11273 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
11275 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
11276 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
11278 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
11280 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
11281 argument is checked for zero before entering loop, avoid checking again.
11283 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
11285 * gimplify.h (generic_expr_could_trap_p): Set as global function.
11286 * gimplify.h (generic_expr_could_trap_p): Likwise.
11287 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
11289 2018-07-06 Jakub Jelinek <jakub@redhat.com>
11291 PR tree-optimization/86401
11292 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
11293 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
11294 (fold_bit_and_mask): ... here. New helper function for match.pd.
11295 * fold-const.h (fold_bit_and_mask): Declare.
11296 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
11298 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
11301 * target.def (translate_mode_attribute): New hook.
11302 * targhooks.h (default_translate_mode_attribute): Declare.
11303 * targhooks.c (default_translate_mode_attribute): New function.
11304 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
11305 * doc/tm.texi: Regenerate.
11306 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
11307 (rs6000_translate_mode_attribute): New function.
11309 2018-07-06 Paul Koning <ni1d@arrl.net>
11311 * doc/md.texi (define_split): Document DONE and FAIL.
11312 (define_peephole2): Ditto.
11314 2018-07-05 Jeff Law <law@redhat.com>
11316 PR tree-optimization/86010
11317 * tree-ssa-dse.c (compute_trims): More aggressively trim at
11318 both the head and tail of mem* and str* calls.
11320 2018-07-05 Jim Wilson <jimw@sifive.com>
11322 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
11324 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
11326 * config/aarch64/aarch64-simd.md: correct flags text for
11327 MIN_EXPR replacement.
11329 2018-07-05 James Clarke <jrtc27@jrtc27.com>
11331 * configure: Regenerated.
11333 2018-07-05 Carl Love <cel@us.ibm.com>
11335 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
11336 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
11337 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
11338 VSX_BUILTIN_DOUBLEL_V4SF.
11340 2018-07-05 Martin Sebor <msebor@redhat.com>
11343 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
11344 than its domain to compute its the upper bound of a char array.
11346 2018-07-05 Nathan Sidwell <nathan@acm.org>
11348 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
11349 * doc/cpp.texi: Update comment.
11350 * doc/tm.texi: Rebuilt.
11351 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
11352 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
11353 * doc/extend.texi (Backwards Compatibility): Clarify it is system
11354 headers affected by extern "C".
11355 * system.h: Poison NO_IMPLICIT_EXTERN_C.
11356 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
11357 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
11358 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
11359 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
11360 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
11361 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
11362 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
11363 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
11364 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
11365 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
11366 config/sparc/sp64-elf.h, config/spu/spu.h,
11367 config/stormy16/stormy16.h, config/v850/v850.h,
11368 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
11369 define NO_IMPLICIT_EXTERN_C.
11370 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
11372 2018-07-05 Tamar Christina <tamar.christina@arm.com>
11375 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
11376 instead of GET_MODE_SIZE when comparing Units.
11378 2018-07-05 Tamar Christina <tamar.christina@arm.com>
11381 * rtlanal.c (set_noop_p): Constrain on mode change,
11382 include hard-reg-set.h
11384 2018-07-05 Tamar Christina <tamar.christina@arm.com>
11386 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
11388 2018-07-05 Jakub Jelinek <jakub@redhat.com>
11391 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
11394 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
11396 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
11398 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
11401 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
11403 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
11405 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
11407 PR middle-end/86380
11408 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
11410 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
11412 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
11413 neg_*_op* variables.
11415 2018-07-04 Martin Liska <mliska@suse.cz>
11417 * tree-switch-conversion.c: Define
11418 max_ratio_for_speed and max_ratio_for_size constants.
11420 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
11421 Martin Liska <mliska@suse.cz>
11423 PR middle-end/66240
11426 * common.opt: Rename align options with 'str_' prefix.
11427 * common/config/i386/i386-common.c (set_malign_value): New
11429 (ix86_handle_option): Use it to set -falign-* options/
11430 * config/aarch64/aarch64-protos.h (struct tune_params): Change
11431 type from int to string.
11432 * config/aarch64/aarch64.c: Update default values from int
11434 * config/alpha/alpha.c (alpha_override_options_after_change):
11436 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
11437 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
11438 max skip conditionally.
11439 * config/i386/freebsd.h (SUBALIGN_LOG): New.
11440 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
11441 max skip conditionally.
11442 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
11443 max skip conditionally.
11444 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
11445 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
11446 max skip conditionally.
11447 * config/i386/i386.c (struct ptt): Change type from int to
11449 (ix86_default_align): Set default values.
11450 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
11451 max skip conditionally.
11452 * config/i386/iamcu.h (SUBALIGN_LOG): New.
11453 (ASM_OUTPUT_MAX_SKIP_ALIGN):
11454 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
11455 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
11456 max skip conditionally.
11457 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
11458 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
11459 * config/i386/x86-64.h (SUBALIGN_LOG): New.
11460 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
11461 max skip conditionally.
11462 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
11463 * config/ia64/ia64.c (ia64_option_override): Set default values
11464 for alignment options.
11465 * config/m68k/m68k.c: Handle new str_align_* options.
11466 * config/mips/mips.c (mips_set_compression_mode): Change
11468 (mips_option_override): Set default values for options.
11469 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
11471 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11473 * config/rx/rx.c (rx_option_override): Likewise.
11474 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
11475 (LABEL_ALIGN): Use align_labels_log.
11476 (LOOP_ALIGN): Use align_loops_align.
11477 * config/s390/s390.c (s390_asm_output_function_label): Use new
11479 * config/sh/sh.c (sh_override_options_after_change):
11480 Change type of constants.
11481 * config/spu/spu.c (spu_sched_init): Likewise.
11482 * config/sparc/sparc.c (sparc_option_override): Set default
11483 values for options.
11484 * config/visium/visium.c (visium_option_override): Likewise.
11485 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
11486 emit p2align format with last argument if it's not needed.
11487 * doc/invoke.texi: Document extended format of -falign-*.
11488 * final.c: Use align_labels alignment.
11489 * flags.h (struct target_flag_state): Change type to use
11491 (struct align_flags_tuple): New.
11492 (struct align_flags): Likewise.
11493 (align_loops_log): Redefine macro to use new types.
11494 (align_loops_max_skip): Redefine macro to use new types.
11495 (align_jumps_log): Redefine macro to use new types.
11496 (align_jumps_max_skip): Redefine macro to use new types.
11497 (align_labels_log): Redefine macro to use new types.
11498 (align_labels_max_skip): Redefine macro to use new types.
11499 (align_functions_log): Redefine macro to use new types.
11500 (align_loops): Redefine macro to use new types.
11501 (align_jumps): Redefine macro to use new types.
11502 (align_labels): Redefine macro to use new types.
11503 (align_functions): Redefine macro to use new types.
11504 (align_functions_max_skip): Redefine macro to use new types.
11505 (align_loops_value): New macro.
11506 (align_jumps_value): New macro.
11507 (align_labels_value): New macro.
11508 (align_functions_value): New macro.
11509 * function.c (invoke_set_current_function_hook): Propagate
11510 alignment values from flags to global variables default in
11512 * ipa-icf.c (sem_function::equals_wpa): Use
11513 cl_optimization_option_eq instead of memcmp.
11514 * lto-streamer.h (cl_optimization_stream_out): Support streaming
11516 (cl_optimization_stream_in): Likewise.
11517 * optc-save-gen.awk: Support strings in cl_optimization.
11518 * opth-gen.awk: Likewise.
11519 * opts.c (finish_options): Remove error checking of invalid
11521 (MAX_CODE_ALIGN): Remove.
11522 (MAX_CODE_ALIGN_VALUE): Likewise.
11523 (parse_and_check_align_values): New function.
11524 (check_alignment_argument): Likewise.
11525 (common_handle_option): Use check_alignment_argument.
11526 * opts.h (parse_and_check_align_values): Declare.
11527 * toplev.c (init_alignments): Remove.
11528 (read_log_maxskip): New.
11529 (parse_N_M): Likewise.
11530 (parse_alignment_opts): Likewise.
11531 (backend_init_target): Remove usage of init_alignments.
11532 * toplev.h (parse_alignment_opts): Declare.
11533 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
11535 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
11536 * tree.c (cl_option_hasher::equal): New.
11537 * varasm.c: Use new global macros.
11539 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
11541 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
11542 Use a simpler align directive also if MAXSKIP = ALIGN-1.
11543 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11544 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11545 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11546 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
11547 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
11548 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
11549 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
11550 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11551 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11552 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11553 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11555 2018-07-04 Martin Liska <mliska@suse.cz>
11556 Jonathan Wakely <jwakely@redhat.com>
11558 * coverage.c: Use correct type.
11559 * doc/invoke.texi: Language correction.
11561 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
11564 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
11565 ENDBRANCH for non-tail call which may return via indirect branch.
11566 * doc/extend.texi: Document indirect_return attribute.
11568 2018-07-03 Martin Sebor <msebor@redhat.com>
11570 PR tree-optimization/86274
11571 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
11573 (format_floating): Correct handling of infinities and NaNs.
11575 2018-07-03 Martin Sebor <msebor@redhat.com>
11577 * print-tree.c (print_real_cst): New function.
11578 (print_node_brief): Call it.
11579 (print_node): Ditto.
11581 2018-07-03 Jeff Law <law@redhat.com>
11583 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
11584 into a single pattern.
11586 * config/h8300/h8300.md (ors code_iterator): New.
11587 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
11588 a single pattern and single splitter.
11589 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
11590 (iorqi3_1, xorqi3_1): Likewise.
11591 (iorqi3, xorqi3 expanders): Similarly.
11593 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
11594 (movmd_internal) into a single pattern using the P mode iterator.
11595 (movmd splitters): Similarly.
11596 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
11597 (movsd splitters): Similarly.
11599 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
11600 ADDB, ADDW and ADDL into a single ADD attribute which selects the
11601 right table based on the size of the operand.
11602 * config/h8300/h8300.md (length_table): Corresponding changes. All
11603 references to "addb", "addw" and "addl" changed to "add".
11604 (btst patterns): Merge two variants into a single pattern.
11605 (tstqi, tsthi): Likewise.
11606 (addhi3_incdec, addsi3_incdec): Likewise.
11607 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
11608 (mulhi3, mulsi3): Likewise.
11609 (udivhi3, udivsi3): Likewise.
11610 (divhi3, divsi3): Likewise.
11611 (andorqi3, andorhi3, andorsi3): Likewise.
11613 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
11616 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
11617 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
11619 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11621 PR tree-optimization/85694
11622 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
11623 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
11625 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
11626 (<u>avg<mode>3_ceil): New patterns.
11628 2018-07-03 David Malcolm <dmalcolm@redhat.com>
11630 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
11631 scan-tree-dump directive.
11632 * gcc.dg/vect/slp-perm-2.c: Likewise.
11633 * gcc.dg/vect/slp-perm-3.c: Likewise.
11634 * gcc.dg/vect/slp-perm-5.c: Likewise.
11635 * gcc.dg/vect/slp-perm-6.c: Likewise.
11636 * gcc.dg/vect/slp-perm-7.c: Likewise.
11637 * gcc.dg/vect/slp-perm-8.c: Likewise.
11639 2018-07-03 Marek Polacek <polacek@redhat.com>
11641 PR middle-end/86202
11642 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
11644 2018-07-03 Richard Biener <rguenther@suse.de>
11647 * tree-ssa-structalias.c (find_func_clobbers): Properly
11648 handle indirect calls.
11650 2018-07-03 Jeff Law <law@redhat.com>
11652 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
11653 (shifts): New code iterator.
11654 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
11655 expander. Fix HImode handling on H8/SX.
11656 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
11657 (subqi3, subhi3, subsi3 expanders): Likewise.
11658 (andqi3, andhi3, andsi3 expanders): Likewise.
11659 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
11660 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
11661 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
11662 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
11663 (zero_extendqihi2, zero_extendqisi2): Likewise.
11664 (extendqihi2, extendqisi2): Likewise.
11665 (rotlqi3, rotlhi3, rotlsi3): Likewise.
11666 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
11667 (rotlqi3_1, rotlhi3_1): Likewise.
11668 (logicalhi3_sn, logicalsi3_sn): Likewise.
11669 (logicalhi3, logicalsi3): Likewise.
11671 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11673 * tree-vect-patterns.c (vect_recog_rotate_pattern)
11674 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
11675 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
11676 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
11677 type to append_pattern_def_seq instead of creating a stmt_vec_info
11679 (build_mask_conversion): Likewise. Remove vinfo argument.
11680 (vect_add_conversion_to_patterm): Likewise, renaming to...
11681 (vect_add_conversion_to_pattern): ...this.
11682 (vect_recog_mask_conversion_pattern): Update call to
11683 build_mask_conversion. Pass the vector type to
11684 append_pattern_def_seq here too.
11685 (vect_recog_gather_scatter_pattern): Update call to
11686 vect_add_conversion_to_pattern.
11688 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11690 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
11691 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
11692 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
11693 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
11694 STMT_VINFO_PATTERN_DEF_SEQ to null here.
11695 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
11696 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
11697 append_pattern_def_seq instead of new_pattern_def_seq.
11698 (vect_recog_divmod_pattern): Do both of the above.
11699 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
11702 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11704 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
11705 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
11706 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
11707 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
11708 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
11709 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
11710 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
11711 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
11712 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
11713 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
11714 parameter with a single stmt_vec_info.
11715 (vect_recog_func_ptr): Likewise.
11716 (vect_recog_gather_scatter_pattern): Likewise, folding in...
11717 (vect_try_gather_scatter_pattern): ...this.
11718 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
11719 the stmt_vec_info of the statement to be matched. Don't clear
11720 STMT_VINFO_RELATED_STMT.
11721 (vect_pattern_recog): Update call accordingly.
11723 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11725 PR tree-optimization/85694
11726 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
11727 (uavgM3_ceil): Document new optabs.
11728 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
11729 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
11731 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
11732 (savg_ceil_optab): New optabs.
11733 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
11734 (vect_vect_recog_func_ptrs): Add it.
11735 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
11736 constant directly from the associated lhs.
11738 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11740 * tree-vect-patterns.c (vect_split_statement): New function.
11741 (vect_convert_input): Use it to try to split an existing cast.
11743 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11745 * poly-int.h (print_hex): New function.
11746 * dumpfile.h (dump_dec, dump_hex): Declare.
11747 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
11748 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
11749 min_input_precision, operation_precision and operation_sign.
11750 * tree-vect-patterns.c (vect_get_range_info): New function.
11751 (vect_same_loop_or_bb_p, vect_single_imm_use)
11752 (vect_operation_fits_smaller_type): Delete.
11753 (vect_look_through_possible_promotion): Add an optional
11754 single_use_p parameter.
11755 (vect_recog_over_widening_pattern): Rewrite to use new
11756 stmt_vec_info infomration. Handle one operation at a time.
11757 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
11758 (vect_truncatable_operation_p, vect_set_operation_type)
11759 (vect_set_min_input_precision): New functions.
11760 (vect_determine_min_output_precision_1): Likewise.
11761 (vect_determine_min_output_precision): Likewise.
11762 (vect_determine_precisions_from_range): Likewise.
11763 (vect_determine_precisions_from_users): Likewise.
11764 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
11765 (vect_vect_recog_func_ptrs): Put over_widening first.
11767 (vect_pattern_recog): Call vect_determine_precisions.
11769 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11771 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
11772 statements that have been replaced by further pattern statements.
11773 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
11775 2018-07-03 Richard Biener <rguenther@suse.de>
11777 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
11778 always set *dt. Dump vectype in vectype overload.
11779 * dumpfile.h (dump_gimple_expr): New function.
11780 (dump_gimple_expr_loc): Likewise.
11781 * dumpfile.c (dump_gimple_expr): New function.
11782 (dump_gimple_expr_loc): Likewise.
11784 2018-07-02 Jeff Law <law@redhat.com>
11786 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
11787 the H8/300, H8/300H and H8/S variants into a single pattern.
11788 (movhi_h8300, movqi_h8300hs): Similarly.
11789 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
11790 (QHI mode iterator): New.
11792 * config/h8300/h8300.md: Remove trailing whitespace.
11794 2018-07-02 Jim Wilson <jimw@sifive.com>
11796 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
11797 instead of emit_insn for interrupt returns.
11798 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
11799 (riscv_sret, riscv_uret): Likewise.
11801 2018-07-02 David Malcolm <dmalcolm@redhat.com>
11803 * pretty-print.c (selftest::test_pp_format): Move save and restore
11804 of quotes to class auto_fix_quotes, and add an instance.
11805 * selftest.c: Include "intl.h".
11806 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
11807 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
11808 * selftest.h (selftest::auto_fix_quotes): New class.
11810 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
11812 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
11813 (aarch64_sve_prepare_conditional_op): Remove.
11814 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
11815 Allow aarch64_simd_reg_or_zero as select operand; remove
11816 the aarch64_sve_prepare_conditional_op call.
11817 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
11818 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
11819 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
11820 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
11821 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
11822 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
11823 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
11824 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
11825 and a splitters to match all of the *_any patterns.
11826 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
11828 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
11829 (SVE_COND_FP_BINARY_REV): Remove.
11830 (sve_int_op_rev, sve_fp_op_rev): New.
11831 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
11832 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
11833 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
11834 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
11835 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
11836 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
11837 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
11838 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
11839 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
11841 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
11842 Remove match_dup 1 from the inner unspec.
11843 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
11845 * config/aarch64/aarch64.md (movprfx): New attr.
11846 (length): Default movprfx to 8.
11847 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
11848 (*madd<SVE_I>, *msub<SVE_I): Likewise.
11849 (*<su>mul<SVE_I>3_highpart): Likewise.
11850 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
11851 (*v<ASHIFT><SVE_I>3): Likewise.
11852 (*<su><MAXMIN><SVE_I>3): Likewise.
11853 (*<su><MAXMIN><SVE_F>3): Likewise.
11854 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
11855 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
11856 (*div<SVE_F>4): Likewise.
11858 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
11860 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
11863 2018-07-02 Richard Biener <rguenther@suse.de>
11865 PR tree-optimization/86363
11866 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
11867 memset argument refers to a non-variable address.
11869 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
11871 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
11872 {PLUS,MINUS}_EXPR code to...
11873 (adjust_symbolic_bound): ...here,
11874 (combine_bound): ...here,
11875 (set_value_range_with_overflow): ...and here.
11877 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
11879 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
11881 (extract_range_from_abs_expr): ...here.
11883 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
11885 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
11886 -fno-omit-frame-pointer when not optimizing.
11888 2018-07-02 Martin Liska <mliska@suse.cz>
11891 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
11892 (propagate_nothrow): Likewise.
11894 2018-07-02 Martin Liska <mliska@suse.cz>
11897 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
11899 2018-07-02 David Malcolm <dmalcolm@redhat.com>
11901 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
11902 function in r262149, changing "loc" param from source_location to
11903 const dump_location_t &.
11904 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
11905 declaration, as above.
11907 2018-07-01 Paul Koning <ni1d@arrl.net>
11909 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
11910 -munit-asm, -mgnu-asm, -mdec-asm.
11911 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
11912 (pdp11_output_labelref): New.
11913 (pdp11_output_def): New.
11914 (pdp11_output_addr_vec_elt): New.
11915 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
11916 %# and %@ format codes.
11917 (pdp11_option_override): New.
11918 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
11919 (pdp11_output_ident): New.
11920 (pdp11_asm_named_section): New.
11921 (pdp11_asm_init_sections): New.
11922 (pdp11_file_start): New.
11923 (pdp11_file_end): New.
11924 (output_ascii): Use .ascii/.asciz for -mdec-asm.
11925 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
11926 %o, like %c but octal.
11927 (pdp11_option_override): New.
11928 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
11930 (DATA_SECTION_ASM_OP): Ditto.
11931 (READONLY_DATA_SECTION_ASM_OP): New.
11932 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
11933 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
11934 (ASM_OUTPUT_LABELREF): Ditto.
11935 (ASM_OUTPUT_DEF): Ditto.
11936 (ASM_OUTPUT_EXTERNAL): New.
11937 (ASM_OUTPUT_SOURCE_FILENAME): New.
11938 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
11939 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
11940 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
11941 %# and %@ format codes.
11942 * config/pdp11/pdp11.opt (mgnu-asm): New.
11943 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
11944 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
11945 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
11947 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
11949 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
11950 dereferencing path[] beyond its length.
11952 (debug_all_paths): New.
11953 (rewire_first_differing_edge): New.
11954 (adjust_paths_after_duplication): New.
11955 (duplicate_thread_path): Call adjust_paths_after_duplication.
11957 (thread_through_all_blocks): Add new argument to
11958 duplicate_thread_path.
11960 2018-06-30 Jim Wilson <jimw@sifive.com>
11962 * config/riscv/predicates.md (p2m1_shift_operand): New.
11963 (high_mask_shift_operand): New.
11964 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
11965 pattern using p2m1_shift_operand.
11966 (lshsi3_zero_extend_3+2): New combiner pattern using
11967 high_mask_shift_operand.
11969 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11971 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
11973 (vect_recog_rotate_pattern): ...here.
11974 (vect_convert_input): Try to insert casts of invariants in the
11976 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
11977 preheader to be empty.
11979 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11981 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
11982 vector type. If given, install it in the new statement's
11983 STMT_VINFO_VECTYPE.
11984 (vect_element_precision): New function.
11985 (vect_unpromoted_value): New struct.
11986 (vect_unpromoted_value::vect_unpromoted_value): New function.
11987 (vect_unpromoted_value::set_op): Likewise.
11988 (vect_look_through_possible_promotion): Likewise.
11989 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
11990 (vect_widened_op_tree, vect_convert_input): Likewise.
11991 (vect_convert_inputs, vect_convert_output): Likewise.
11992 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
11993 to handle the optional cast of the multiplication result and
11994 vect_widened_op_tree to detect the widened multiplication itself.
11995 Do not require the input and output of promotion casts to have
11996 the same sign, but base the signedness of the operation on the
11997 input rather than the result. If the pattern includes two
11998 promotions, check that those promotions have the same sign.
11999 Do not restrict the MULT_EXPR handling to a double-width result;
12000 handle quadruple-width results and wider. Use vect_convert_inputs
12001 to convert the inputs to the common type.
12002 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
12003 to handle the optional cast of the ABS result. Also allow a sign
12004 change or a sign extension between the ABS and MINUS.
12005 Use vect_widened_op_tree to detect the widened subtraction and use
12006 vect_convert_inputs to convert the inputs to the common type.
12007 (vect_handle_widen_op_by_const): Delete.
12008 (vect_recog_widen_op_pattern): New function.
12009 (vect_recog_widen_mult_pattern): Use it.
12010 (vect_recog_widen_shift_pattern): Likewise.
12011 (vect_recog_widen_sum_pattern): Use
12012 vect_look_through_possible_promotion to handle the promoted
12015 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
12017 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
12018 the containing gimple_seq *.
12019 * gimple-iterator.h (gsi_for_stmt): Declare it.
12020 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
12021 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
12022 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
12023 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
12024 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
12026 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
12028 (vect_mark_pattern_stmts): ...here. Handle cases in which the
12029 statement being replaced is part of an existing pattern
12030 definition sequence, inserting the new pattern statements before
12032 (vect_pattern_recog_1): Don't return a bool. If the statement
12033 is already part of a pattern, instead apply pattern matching
12034 to the pattern definition statements. Don't clear the
12035 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
12036 (vect_pattern_recog): Don't break after the first match;
12037 continue processing the pattern definition statements instead.
12038 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
12040 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
12042 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
12043 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
12044 (vect_recog_widen_sum_pattern): Use it.
12046 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
12048 * tree-vect-loop.c (vectorizable_reduction): Assert that the
12049 phi is not a pattern statement and has not been replaced by
12050 a pattern statement.
12051 * tree-vect-patterns.c (type_conversion_p): Don't check
12052 STMT_VINFO_IN_PATTERN_P.
12053 (vect_recog_vector_vector_shift_pattern): Likewise.
12054 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
12055 the pattern statement rather than the original statement; check
12056 directly for a WIDEN_MULT_EXPR here.
12057 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
12058 vect_is_simple_use to return the pattern statement rather
12059 than the original statement; use is_pattern_stmt_p to check
12060 for such a pattern statement.
12061 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
12062 to return the pattern statement rather than the original statement;
12063 don't do the same transformation here.
12064 (vect_is_simple_use): If the defining statement has been replaced
12065 by a pattern statement, return the pattern statement instead.
12066 Remove the corresponding (local) transformation from the vectype
12069 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
12071 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
12072 end and default to null.
12073 * tree-vect-loop.c (vect_create_epilog_for_reduction)
12074 (vectorizable_reduction): Update calls accordingly, dropping the
12075 gimple ** argument if the passed-back statement isn't needed.
12076 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
12077 (vect_recog_rotate_pattern): Likewise.
12078 (vect_recog_mask_conversion_pattern): Likewise.
12079 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
12080 (vect_mask_constant_operand_p): Likewise.
12081 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
12082 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
12083 (get_group_load_store_type, get_load_store_type): Likewise.
12084 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
12085 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
12086 (vectorizable_conversion, vectorizable_assignment): Likewise.
12087 (vectorizable_shift, vectorizable_operation): Likewise.
12088 (vectorizable_store, vect_is_simple_cond): Likewise.
12089 (vectorizable_condition, vectorizable_comparison): Likewise.
12090 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
12091 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
12092 and move it to the end. Cope with null def_stmt_outs.
12094 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
12096 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
12098 2018-06-29 Jeff Law <law@redhat.com>
12100 * config/v850/v850.c (v850_legitimate_address_p): Handle large
12101 displacements for TARGET_V850E2V3 and newer.
12102 (TARGET_LRA_P): Remove. Defaults to LRA now.
12103 * config/v850/v850.md (sign23byte_load): Remove.
12104 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
12105 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
12107 2018-06-29 Martin Liska <mliska@suse.cz>
12110 * coverage.c (coverage_init): Mangle full path name.
12111 * doc/invoke.texi: Document the change.
12112 * gcov-io.c (mangle_path): New.
12113 * gcov-io.h (mangle_path): Likewise.
12114 * gcov.c (mangle_name): Use mangle_path for path mangling.
12116 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12118 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
12119 if starting source register is not even.
12121 2018-06-29 Martin Liska <mliska@suse.cz>
12123 PR tree-optimization/86263
12124 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
12125 Make edge redirection.
12127 2018-06-29 David Malcolm <dmalcolm@redhat.com>
12129 * dumpfile.c (dump_loc): Add indentation based on scope depth.
12130 (dump_scope_depth): New variable.
12131 (get_dump_scope_depth): New function.
12132 (dump_begin_scope): New function.
12133 (dump_end_scope): New function.
12134 * dumpfile.h (get_dump_scope_depth): New declaration.
12135 (dump_begin_scope): New declaration.
12136 (dump_end_scope): New declaration.
12137 (class auto_dump_scope): New class.
12138 (AUTO_DUMP_SCOPE): New macro.
12139 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
12142 2018-06-29 Richard Biener <rguenther@suse.de>
12144 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
12145 compute_all_dependences succeeds.
12146 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
12147 exceed --param loop-max-datarefs-for-datadeps.
12149 2018-06-29 Jakub Jelinek <jakub@redhat.com>
12151 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
12153 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
12156 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
12157 alternative 0 in preferred_for_speed attribute.
12159 2018-06-28 Paul Koning <ni1d@arrl.net>
12161 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
12162 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
12163 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
12164 * config/pdp11/pdp11.md: Correct "length" attribute calculation
12165 for shift insn patterns.
12167 2018-06-28 David Malcolm <dmalcolm@redhat.com>
12169 * cgraph.c (cgraph_node::get_body): Replace assignments to
12170 "dump_file" with calls to set_dump_file.
12171 * dumpfile.c (alt_dump_file): Make static, and group with...
12172 (alt_flags): ...this definition.
12173 (dumps_are_enabled): New variable.
12174 (refresh_dumps_are_enabled): New function.
12175 (set_dump_file): New function.
12176 (set_alt_dump_file): New function.
12177 (gcc::dump_manager::dump_start): Replace assignments to
12178 "dump_file" and "alt_dump_file" with calls to set_dump_file and
12180 (gcc::dump_manager::dump_finish): Likewise.
12181 * dumpfile.h (alt_dump_file): Delete decl.
12182 (dumps_are_enabled): New variable decl.
12183 (set_dump_file): New function decl.
12184 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
12186 * tree-nested.c (lower_nested_functions): Replace assignments to
12187 "dump_file" with calls to set_dump_file.
12189 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
12191 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
12192 goto_locus of each outgoing edge of each basic block.
12194 2018-06-28 Richard Biener <rguenther@suse.de>
12196 * dwarf2out.c (decl_scope_table): Remove.
12197 (push_decl_scope): Likewise.
12198 (pop_decl_scope): Likewise.
12199 (gen_type_die_for_member): Do not call push/pop_decl_scope.
12200 (gen_struct_or_union_type_die): Likewise.
12201 (gen_tagged_type_die): Likewise.
12202 (dwarf2out_init): Do not initialize decl_scope_table.
12203 (dwarf2out_c_finalize): Do not free it.
12205 2018-06-28 Richard Biener <rguenther@suse.de>
12207 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
12208 deciding whether to not re-use a DIE.
12210 2018-06-28 Richard Biener <rguenther@suse.de>
12212 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
12213 DW_AT_abstract_origin attribute.
12215 2018-06-28 Martin Liska <mliska@suse.cz>
12217 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
12218 Use newly introduced constants.
12219 * tree-switch-conversion.h (struct jump_table_cluster):
12220 Define max_ratio_for_size and max_ratio_for_speed.
12222 2018-06-28 Martin Liska <mliska@suse.cz>
12224 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
12225 Add new checking assert to catch invalid state.
12226 (jump_table_cluster::can_be_handled): Handle single case
12228 (jump_table_cluster::is_beneficial): Bail out for such case.
12229 (bit_test_cluster::find_bit_tests):
12230 Add new checking assert to catch invalid state.
12231 (bit_test_cluster::can_be_handled): Handle single case
12233 (bit_test_cluster::is_beneficial): Bail out for such case.
12234 (switch_decision_tree::analyze_switch_statement):
12237 2018-06-28 Martin Liska <mliska@suse.cz>
12239 * common.opt: Introduce -completion option.
12240 * gcc.c (driver_handle_option): Handle it.
12241 (driver::main): Print completions if completion
12243 * opt-suggestions.c (option_proposer::get_completions):
12245 (option_proposer::suggest_completion): Likewise.
12246 (option_proposer::find_param_completions): Likewise.
12247 (verify_autocompletions): Likewise.
12248 (test_completion_valid_options): Likewise.
12249 (test_completion_valid_params): Likewise.
12250 (in_completion_p): Likewise.
12251 (empty_completion_p): Likewise.
12252 (test_completion_partial_match): Likewise.
12253 (test_completion_garbage): Likewise.
12254 (opt_proposer_c_tests): Likewise.
12255 * opt-suggestions.h: Declare new functions.
12256 * opts.c (common_handle_option): Handle OPT__completion_.
12257 * selftest-run-tests.c (selftest::run_tests): Add
12258 opt_proposer_c_tests.
12259 * selftest.c (assert_str_startswith): New.
12260 * selftest.h (assert_str_startswith): Likewise.
12261 (opt_proposer_c_tests): New.
12262 (ASSERT_STR_STARTSWITH): Likewise.
12264 2018-06-28 Martin Liska <mliska@suse.cz>
12266 * Makefile.in: Add opt-suggestions.o.
12267 * gcc-main.c: Include opt-suggestions.h.
12268 * gcc.c (driver::driver): Likewise.
12269 (driver::~driver): Remove m_option_suggestions.
12270 (driver::build_option_suggestions): Moved to option_proposer.
12271 (driver::suggest_option): Likewise.
12272 (driver::handle_unrecognized_options): Use option_proposer.
12273 * gcc.h (class driver): Add new memver m_option_proposer.
12274 * opt-suggestions.c: New file.
12275 * opt-suggestions.h: New file.
12277 2018-06-28 Martin Liska <mliska@suse.cz>
12279 * vec.h (class auto_string_vec): New (moved from auto_argvec).
12280 (auto_string_vec::~auto_string_vec): Likewise.
12282 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
12284 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
12285 prevent_decl_creation_for_types fields up and add reset_location field.
12286 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
12287 statement if id->reset_location is true.
12288 (copy_edges_for_bb): Do not set goto_locus on the new edges if
12289 id->reset_location is true.
12290 (copy_phis_for_bb): Force input_location on the arguments if
12291 id->reset_location is true.
12292 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
12293 is set on the function to be inlined.
12295 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
12297 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
12299 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
12301 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
12302 registers for Pmode.
12303 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
12304 hard registers for the clobbered pseudo.
12306 2018-06-27 Paul Koning <ni1d@arrl.net>
12308 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
12309 mutually exclusive options.
12310 * config/pdp11/constraints.md (h): New constraint.
12311 (O): Update definition to match shift code generation.
12312 (D): New constraint.
12313 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
12315 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
12317 (output_jump): Change arguments.
12318 (pdp11_fixed_cc_regs): New function.
12319 (pdp11_cc_mode): Ditto.
12320 (pdp11_expand_shift): Ditto.
12321 (pdp11_assemble_shift): Ditto.
12322 (pdp11_small_shift): Ditto.
12323 (pdp11_branch_cost): Remove.
12324 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
12326 (pdp11_register_move_cost): Update for CC registers.
12327 (pdp11_rtx_costs): Add case for LSHIFTRT.
12328 (pdp11_output_jump): Add CCNZ mode conditional branches.
12329 (notice_update_cc_on_set): Remove.
12330 (pdp11_cc_mode): New function.
12331 (simple_memory_operand): Correct pre/post decrement case.
12332 (no_side_effect_operand): New function.
12333 (pdp11_regno_reg_class): Add CC_REGS class.
12334 (pdp11_fixed_cc_regs): New function.
12335 (pdp11_small_shift): New function.
12336 (pdp11_expand_shift): New function to expand shift insns.
12337 (pdp11_assemble_shift): New function to output shifts.
12338 (pdp11_branch_cost): Remove.
12339 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
12340 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
12341 (WCHAR_TYPE): Ditto.
12342 (PTRDIFF_TYPE): Ditto.
12343 (ADJUST_INSN_LENGTH): New macro.
12344 (FIXED_REGISTERS): Add CC registers.
12345 (CALL_USED_REGISTERS): Ditto.
12346 (reg_class): Ditto.
12347 (REG_CLASS_NAMES): Ditto.
12348 (REG_CLASS_CONTENTS): Ditto.
12349 (SELECT_CC_MODE): Use new function.
12350 (TARGET_FLAGS_REGNUM): New macro.
12351 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
12352 (cc0_reg_rtx): Remove.
12353 (CC_STATUS_MDEP): Remove.
12354 (CC_STATUS_MDEFP_INIT): Remove.
12355 (CC_IN_FPU): Remove.
12356 (NOTICE_UPDATE_CC): Remove.
12357 (REGISTER_NAMES): Add CC registers.
12358 (BRANCH_COST): Change to constant 1.
12359 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
12361 * config/pdp11/pdp11.opt (mbcopy): Remove.
12362 (mbcopy-builtin): Remove.
12363 (mbranch-cheap): Remove.
12364 (mbranch-expensive): Remove.
12365 * config/pdp11/predicates.md (expand_shift_operand): Update to
12366 match shift code generation.
12367 (ccnz_operator): New predicate.
12368 * doc/invoke.texi (PDP-11 Options): Remove deleted options
12369 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
12370 Remove non-existent option -mabshi, -mno-abshi. Document mutually
12372 * doc/md.texi (PDP-11): Document new D and h constraints. Update
12373 description of O constraint.
12375 2018-06-27 Jeff Law <law@redhat.com>
12376 Austin Law <austinklaw@gmail.com>
12378 * config/v850/v850.md (addsi3_set_flags): New pattern.
12379 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
12380 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
12381 (zero_extendhisi2_v850_set_flags): Likewise.
12382 (zero_extendqisi2_v850_set_flags): Likewise.
12383 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
12384 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
12385 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
12387 * config/v850/v850-protos.h (notice_update_cc): Remove.
12388 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
12389 (v850_print_operand): Handle 'D' and "d".
12390 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
12391 Add handling of arithmetic/logical operations compared against zero.
12392 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
12393 Do not look at v850_compare_op, instead get mode from last argument.
12394 (v850_gen_compare): Remove
12395 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
12396 after reload for prologue insns.
12397 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
12399 (construct_save_jarl): Likewise.
12400 (TARGET_FLAGS_REGNUM): Define.
12401 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
12402 (NOTICE_UPDATE_CC): Remove.
12403 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
12404 than cc0. Conditionalize on reload_completed.
12405 (cmpsi_insn, setfcc_insn): Likewise.
12406 (tst1 splitter): Turn into define_and_split which sets the flags
12408 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
12409 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
12410 (cstoresf4, cstoredf4): Clobber the flags.
12411 (cmpsi, cmpsf, cmpdf): Remove expanders.
12412 (setf_insn): Remove pattern.
12413 (addsi3): Turn into define_and_split which clobbers the flags after
12414 reload and a suitable pattern (addsi3_clobber_flags) for use after
12416 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
12417 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
12418 (ashrsi3, ashrsi3_v850e2): Likewise.
12419 (bins): Clobber the flags.
12420 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
12421 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
12422 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
12423 (call_value_internal_short, call_value_internal_long): Likewise.
12424 (callt_save_interrupt, callt_return_interrupt): Likewise.
12425 (save_interrupt, return_interrupt): Likewise.
12426 (callt_save_all_interrupt, save_all_interrupt): Likewise.
12427 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
12428 (restore_all_interrupt, _restore_all_interrupt): Likewise.
12429 (All FP comparisons): Only allow after reload has completed.
12431 (divh, divhu): Tweak output template.
12432 (branch_z_normal, branch_z_invert): Remove
12433 (branch_nz_normal, branch_nz_invert): Likewise.
12434 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
12436 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
12437 * config/v850/v850.c (notice_update_cc): Remove.
12438 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
12439 (CC_NO_CARRY): Likewise.
12440 (NOTICE_UPDATE_CC): Define to nothing.
12441 * config/v850/v850.md: Remove block comment on cc0 handling
12442 Remove "cc" attribute from all patterns. Remove cc_status handling
12443 from all patterns. Minor formatting fixes.
12445 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12447 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
12448 (cortex-a76.cortex-a55): Likewise.
12449 * config/aarch64/aarch64-tune.md: Regenerate.
12450 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
12451 cortex-a76.cortex-a55.
12453 2018-06-27 Jeff Law <law@redhat.com>
12455 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
12456 (MULTILIB_DIRNAMES): Similarly.
12458 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
12460 * gimple.h (gimple_return_retbnd): Delete.
12461 (gimple_return_set_retbnd): Likewise.
12462 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
12463 gimple_return_set_retbnd.
12464 * gimple-pretty-print.c (dump_gimple_return): Remove call to
12465 gimple_return_retbnd and adjust.
12466 * tree-inline.h (struct copy_body_data): Remove retbnd field.
12467 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
12468 Explicitly return NULL in a couple more cases. Move assertion
12469 on debug statements and remove unreachable code.
12470 (reset_debug_binding): Do not test id->retbnd.
12471 (expand_call_inline): Do not set it.
12473 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
12475 * configure.ac: Add --disable-gcov option.
12476 * configure: Regenerate.
12477 * Makefile.in: Honour @enable_gcov@.
12478 * doc/install.texi: Document --disable-gcov.
12480 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12482 * config/arm/arm-cpus.in (cortex-a76): New entry.
12483 (cortex-a76.cortex-a55): Likewise.
12484 * config/arm/arm-tables.opt: Regenerate.
12485 * config/arm/arm-tune.md: Likewise.
12486 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
12487 * doc/invoke.texi (ARM Options): Document cortex-a76 and
12488 cortex-a76.cortex-a55.
12490 2018-06-27 Tamar Christina <tamar.christina@arm.com>
12493 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
12495 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
12497 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
12499 (EPILOGUE_USES): Likewise.
12501 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
12503 * tree-inline.c (remap_location): New function extracted from...
12504 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
12505 (copy_phis_for_bb): ...here. Call remap_location.
12506 (copy_cfg_body): Adjust call to copy_edges_for_bb.
12508 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
12510 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
12511 unaligned vsx for 16B memset.
12513 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
12516 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
12517 ieee128_float_type_node to long_double_type_node unless
12518 TARGET_LONG_DOUBLE_128 is set.
12520 2018-06-26 David Malcolm <dmalcolm@redhat.com>
12522 * cfgloop.c (get_loop_location): Convert return type from
12523 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
12524 by implicit construction from rtx_insn *, and using
12525 dump_user_location_t::from_function_decl for the fallback case.
12526 * cfgloop.h (get_loop_location): Convert return type from
12527 location_t to dump_user_location_t.
12528 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
12529 dump_printf_loc to pass in a dump_location_t rather than a
12530 location_t, via the gimple stmt.
12531 * coverage.c (get_coverage_counts): Update calls to
12532 dump_printf_loc to pass in dump_location_t rather than a
12534 * doc/optinfo.texi (Dump types): Convert example of
12535 dump_printf_loc from taking "locus" to taking "insn". Update
12536 description of the "_loc" calls to cover dump_location_t.
12537 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
12539 (dump_user_location_t::dump_user_location_t): New constructors,
12540 from gimple *stmt and rtx_insn *.
12541 (dump_user_location_t::from_function_decl): New function.
12542 (dump_loc): Make static.
12543 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
12544 const dump_location_t &.
12545 (dump_generic_expr_loc): Delete.
12546 (dump_printf_loc): Convert param "loc" from location_t to
12547 const dump_location_t &.
12548 (selftest::test_impl_location): New function.
12549 (selftest::dumpfile_c_tests): New function.
12550 * dumpfile.h: Include "profile-count.h".
12551 (class dump_user_location_t): New class.
12552 (struct dump_impl_location_t): New struct.
12553 (class dump_location_t): New class.
12554 (dump_printf_loc): Convert 2nd param from source_location to
12555 const dump_location_t &.
12556 (dump_generic_expr_loc): Delete.
12557 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
12558 const dump_location_t &.
12559 * gimple-fold.c (fold_gimple_assign): Update call to
12560 dump_printf_loc to pass in a dump_location_t rather than a
12561 location_t, via the gimple stmt.
12562 (gimple_fold_call): Likewise.
12563 * gimple-loop-interchange.cc
12564 (loop_cand::analyze_iloop_reduction_var): Update for change to
12565 check_reduction_path.
12566 (tree_loop_interchange::interchange): Update for change to
12567 find_loop_location.
12568 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
12569 change in return-type of find_loop_location.
12570 (graphite_regenerate_ast_isl): Likewise.
12571 * graphite-optimize-isl.c (optimize_isl): Likewise.
12572 * graphite.c (graphite_transform_loops): Likewise.
12573 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
12574 pass in a dump_location_t rather than a location_t, via the
12576 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
12577 * ipa.c (walk_polymorphic_call_targets): Likewise.
12578 * loop-unroll.c (report_unroll): Convert "locus" param from
12579 location_t to dump_location_t.
12580 (decide_unrolling): Update for change to get_loop_location's
12582 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
12583 location_t to dump_user_location_t.
12584 (grid_find_single_omp_among_assignments_1): Updates calls to
12585 dump_printf_loc to pass in a dump_location_t rather than a
12586 location_t, via the gimple stmt.
12587 (grid_parallel_clauses_gridifiable): Convert "tloc" from
12588 location_t to dump_location_t. Updates calls to dump_printf_loc
12589 to pass in a dump_location_t rather than a location_t, via the
12591 (grid_inner_loop_gridifiable_p): Likewise.
12592 (grid_dist_follows_simple_pattern): Likewise.
12593 (grid_gfor_follows_tiling_pattern): Likewise.
12594 (grid_target_follows_gridifiable_pattern): Likewise.
12595 (grid_attempt_target_gridification): Convert initialization
12596 of local "grid" from memset to zero-initialization; FIXME: does
12597 this require C++11? Update call to dump_printf_loc to pass in a
12598 optinfo_location rather than a location_t, via the gimple stmt.
12599 * profile.c (read_profile_edge_counts): Updates call to
12600 dump_printf_loc to pass in a dump_location_t rather than a
12602 (compute_branch_probabilities): Likewise.
12603 * selftest-run-tests.c (selftest::run_tests): Call
12605 * selftest.h (dumpfile_c_tests): New decl.
12606 * tree-loop-distribution.c (pass_loop_distribution::execute):
12607 Update for change in return type of find_loop_location.
12608 * tree-parloops.c (parallelize_loops): Likewise.
12609 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
12610 "locus" from location_t to dump_user_location_t.
12611 (canonicalize_loop_induction_variables): Likewise.
12612 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
12613 for change in return type of find_loop_location.
12614 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
12615 to dump_printf_loc to pass in a dump_location_t rather than a
12616 location_t, via the stmt.
12617 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
12619 * tree-vect-loop-manip.c (find_loop_location): Convert return
12620 type from source_location to dump_user_location_t.
12621 (vect_do_peeling): Update for above change.
12622 (vect_loop_versioning): Update for change in type of
12624 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
12625 from location_t to dump_user_location_t.
12626 (vect_estimate_min_profitable_iters): Update for change in type
12628 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
12629 location_t to dump_location_t.
12630 (vect_slp_bb): Update for change in type of vect_location.
12631 * tree-vectorizer.c (vect_location): Convert from source_location
12632 to dump_user_location_t.
12633 (try_vectorize_loop_1): Update for change in vect_location's type.
12634 (vectorize_loops): Likewise.
12635 (increase_alignment): Likewise.
12636 * tree-vectorizer.h (vect_location): Convert from source_location
12637 to dump_user_location_t.
12638 (find_loop_location): Convert return type from source_location to
12639 dump_user_location_t.
12640 (check_reduction_path): Convert 1st param from location_t to
12641 dump_user_location_t.
12642 * value-prof.c (check_counter): Update call to dump_printf_loc to
12643 pass in a dump_user_location_t rather than a location_t; update
12644 call to error_at for change in type of "locus".
12645 (check_ic_target): Update call to dump_printf_loc to
12646 pass in a dump_user_location_t rather than a location_t, via the
12649 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
12651 * config/s390/s390.h (enum processor_flags): Do not use
12652 default tune parameter when -march was specified.
12654 2018-06-26 Jakub Jelinek <jakub@redhat.com>
12657 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
12658 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
12661 2018-06-26 Richard Biener <rguenther@suse.de>
12663 PR tree-optimization/86287
12665 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
12666 (vect_analyze_loop): Initialize n_stmts.
12668 2018-06-26 Richard Biener <rguenther@suse.de>
12670 PR middle-end/86271
12671 * fold-const.c (fold_convertible_p): Pointer extension
12674 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
12677 * dwarf2out.c (loc_list_has_views): Adjust comments.
12678 (dw_loc_list): Split single cross-partition range with
12681 2018-06-25 Jeff Law <law@redhat.com>
12683 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
12684 on -mbig-switch by default.
12686 * config/v850/predicates.md (const_float_1_operand): Fix match_code
12688 (const_float_0_operand): Remove unused predicate.
12689 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
12690 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
12691 (recipsf2): New expander. Original pattern now called
12693 (recipdf2, recipdf2_insn): Similarly.
12694 (rsqrtsf2, rsqrtsf2_insn): Similarly
12695 (rsqrtdf2, rsqrtdf2_insn): Similarly
12697 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
12699 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
12700 Simplify logic for FreeBSD (twice).
12702 2018-06-25 Martin Sebor <msebor@redhat.com>
12704 PR tree-optimization/86204
12705 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
12706 a strnlen result if it's less than the length of the string.
12708 2018-06-25 Martin Sebor <msebor@redhat.com>
12710 PR tree-optimization/85700
12711 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
12712 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
12713 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
12715 2018-06-25 Martin Sebor <msebor@redhat.com>
12717 * doc/extend.texi (Zero-length arrays): Update and clarify.
12719 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
12721 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
12722 added IEEE/IBM long double multilib support on PowerPC little
12723 endian Linux systems.
12724 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
12725 (MULTILIB_DEFAULTS): Likewise.
12726 * config/rs6000/rs6000.c (rs6000_option_override_internal):
12728 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
12729 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
12730 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
12732 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
12734 PR middle-end/86311
12735 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
12736 (REORDER_45): Likewise.
12738 2018-06-25 Jeff Law <law@redhat.com>
12740 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
12741 dividend to 32 bits. Adjust length.
12742 (udivmodhi4): Cleanup output template. Fix length.
12744 2018-06-25 Carl Love <cel@us.ibm.com>
12746 * config/rs6000/vsx.md: Change word selector to prefered location.
12748 2018-06-25 Richard Biener <rguenther@suse.de>
12750 PR tree-optimization/86304
12751 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
12752 epilogue-if-converted loops as well.
12754 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
12756 * lto-section-out.c (lto_begin_section): Do not print section
12757 name for noaddr and unnumbered dumps.
12759 2018-06-25 Richard Biener <rguenther@suse.de>
12761 * tree-vectorizer.h (struct vec_info_shared): New structure
12762 with parts split out from struct vec_info and loop_nest from
12763 struct _loop_vec_info.
12764 (struct vec_info): Adjust accordingly.
12765 (struct _loop_vec_info): Likewise.
12766 (LOOP_VINFO_LOOP_NEST): Adjust.
12767 (LOOP_VINFO_DATAREFS): Likewise.
12768 (LOOP_VINFO_DDRS): Likewise.
12769 (struct _bb_vec_info): Likewise.
12770 (BB_VINFO_DATAREFS): Likewise.
12771 (BB_VINFO_DDRS): Likewise.
12772 (struct _stmt_vec_info): Add dr_aux member.
12773 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
12774 (DR_MISALIGNMENT_UNINITIALIZED): New.
12775 (set_dr_misalignment): Adjust.
12776 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
12777 (vect_analyze_loop): Adjust prototype.
12778 (vect_analyze_loop_form): Likewise.
12779 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
12780 Compute dependences lazily.
12781 (vect_record_base_alignments): Use shared datarefs/ddrs.
12782 (vect_verify_datarefs_alignment): Likewise.
12783 (vect_analyze_data_refs_alignment): Likewise.
12784 (vect_analyze_data_ref_accesses): Likewise.
12785 (vect_analyze_data_refs): Likewise.
12786 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
12787 constructor parameter for shared part.
12788 (vect_analyze_loop_form): Pass in shared part and adjust.
12789 (vect_analyze_loop_2): Pass in storage for the number of
12790 stmts. Move loop nest finding to the caller. Compute
12792 (vect_analyze_loop): Pass in shared part.
12793 (vect_transform_loop): Verify shared datarefs are unchanged.
12794 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
12795 constructor parameter for shared part.
12796 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
12797 (vect_slp_bb): Verify shared datarefs are unchanged before
12799 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
12801 (new_stmt_vec_info): Initialize DR_AUX misalignment to
12802 DR_MISALIGNMENT_UNINITIALIZED.
12803 * tree-vectorizer.c (vec_info::vec_info): Add constructor
12804 parameter for shared part.
12805 (vec_info::~vec_info): Adjust.
12806 (vec_info_shared::vec_info_shared): New.
12807 (vec_info_shared::~vec_info_shared): Likewise.
12808 (vec_info_shared::save_datarefs): Likewise.
12809 (vec_info_shared::check_datarefs): Likewise.
12810 (try_vectorize_loop_1): Construct shared part live for analyses
12811 of a single loop for multiple vector sizes.
12812 * tree-parloops.c (gather_scalar_reductions): Adjust.
12814 2018-06-25 Richard Biener <rguenther@suse.de>
12816 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
12817 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
12818 (vect_analyze_data_refs): Remove similar code from here and
12819 simplify accordingly.
12821 2018-06-25 Richard Biener <rguenther@suse.de>
12823 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
12824 for reverse storage order accesses rather than asserting
12825 they cannot happen here.
12827 2018-06-25 Tom de Vries <tdevries@suse.de>
12830 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
12831 Use data16 instead of .byte for insn prefix.
12833 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
12836 * parser.c (make_char_string_pack): Pass this literal chars
12837 through cpp_interpret_string.
12838 (cp_parser_userdef_numeric_literal): Check the result of
12839 make_char_string_pack.
12841 2018-06-24 Maya Rashish <coypu@sdf.org>
12843 * ginclude/stddef.h: Simplify conditions around avoiding
12844 re-definition of __size_t.
12846 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
12848 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
12849 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
12851 2018-06-22 Maya Rashish <coypu@sdf.org>
12853 * doc/invoke.texi (mno-fancy-math-387): Update for changes
12854 made to OpenBSD and NetBSD through the years.
12856 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
12858 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
12859 behavior of vec_pack (vector double, vector double) to match
12860 behavior of vec_float2 (vector double, vector double).
12862 2018-06-22 Olivier Hainque <hainque@adacore.com>
12864 * gimplify.c (gimplify_function_tree): Prevent creation
12865 of a trampoline for the address of the current function
12866 passed to entry/exit instrumentation hooks.
12868 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
12871 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
12874 2018-06-22 Martin Liska <mliska@suse.cz>
12876 PR tree-optimization/86263
12877 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
12878 Bail out if is_enabled is false.
12879 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
12881 (jump_table_cluster::is_enabled): New function.
12883 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
12885 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
12886 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
12887 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
12888 (lto_input_ts_binfo_tree_pointers): Likewise.
12889 * tree-streamer-out.c (streamer_write_tree_bitfields,
12890 write_ts_binfo_tree_pointers): Likewise.
12891 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
12893 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
12895 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
12897 2018-06-22 Martin Liska <mliska@suse.cz>
12899 * symbol-summary.h (get): Make it pure and inline move
12900 functionality from ::get function.
12901 (get): Remove and inline into ::get and ::get_create.
12902 (get_create): Move code from ::get function.
12904 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12907 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
12908 -x assembler-with-cpp.
12910 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12912 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
12913 _FILE_OFFSET_BITS=64 for C++.
12915 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
12917 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
12918 conversion insn that shows up when pr85657-3.c is compiled using
12919 IEEE 128-bit long double.
12920 (neg<mode>2_internal): Use the correct mode to check whether the
12921 mode is IBM extended.
12922 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
12923 multiply and divide external functions from being created more
12926 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12928 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
12930 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
12931 the edge can be forwarded.
12932 (cfg_layout_merge_blocks): Likewise.
12934 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12936 * except.c (finish_eh_generation): Commit edge insertions only after
12937 the EH edges have been redirected from post-landing to landing pads.
12939 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12941 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
12942 create_tmp_var_for to create the FRAME decl.
12943 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
12945 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12947 * tree-inline.c (copy_edges_for_bb): Minor tweak.
12948 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
12949 debug statement when resetting its value.
12950 (expand_call_inline): Copy the locus of the call onto the assignment
12951 of the return value, if any. Use local variable in more cases.
12953 2018-06-21 Martin Liska <mliska@suse.cz>
12955 * ipa-pure-const.c (propagate_nothrow): Use
12956 funct_state_summaries->get.
12957 (dump_malloc_lattice): Likewise.
12958 (propagate_malloc): Likewise.
12960 2018-06-21 Richard Biener <rguenther@suse.de>
12962 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
12963 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
12964 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
12966 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
12967 BLOCK_ABSTRACT_ORIGIN unconditionally.
12969 2018-06-21 David Malcolm <dmalcolm@redhat.com>
12971 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
12973 * ipa-reference.c (ipa_reference_c_finalize): Delete
12974 ipa_ref_opt_sum_summaries and set it to NULL.
12976 2018-06-21 Tom de Vries <tdevries@suse.de>
12978 PR tree-optimization/85859
12979 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
12980 test with comment from bb_no_side_effects_p.
12982 2018-06-21 Richard Biener <rguenther@suse.de>
12984 PR tree-optimization/86232
12985 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
12986 max for constant niter.
12988 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
12990 * config/aarch64/aarch64-simd.md
12991 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
12993 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
12995 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
12996 Make opernads of the unspec commutative.
12998 2018-06-21 Richard Biener <rguenther@suse.de>
13000 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
13001 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
13002 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
13003 (vect_analyze_data_ref_dependence): Re-order checks to deal with
13005 (vect_record_base_alignments): Do not record base alignment
13006 for gathers or scatters.
13007 (vect_compute_data_ref_alignment): Drop return value that is always
13008 true. Bail out early for gathers or scatters.
13009 (vect_enhance_data_refs_alignment): Bail out early for gathers
13011 (vect_find_same_alignment_drs): Likewise.
13012 (vect_analyze_data_refs_alignment): Remove dead code.
13013 (vect_slp_analyze_and_verify_node_alignment): Likewise.
13014 (vect_analyze_data_refs): For possible gathers or scatters do
13015 not create an alternate DR, just check their possible validity
13016 and mark them. Adjust DECL_NONALIASED handling to not rely
13017 on DR_BASE_ADDRESS.
13018 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
13019 update inits of gathers or scatters.
13020 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
13021 Also copy gather/scatter flag to pattern vinfo.
13023 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
13025 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
13026 behavior of vec_packsu (vector unsigned long long, vector unsigned
13027 long long) to match behavior of vec_packs with same signature.
13029 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
13030 Thomas Schwinge <thomas@codesourcery.com>
13031 Cesar Philippidis <cesar@codesourcery.com>
13033 * gimplify.c (gimplify_scan_omp_clauses): Add support for
13034 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
13035 (gimplify_adjust_omp_clauses): Likewise.
13036 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
13037 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
13038 (gimplify_omp_target_update): Update handling of acc update and
13040 * omp-low.c (install_var_field): Remove unused parameter
13041 base_pointers_restrict.
13042 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
13043 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
13045 (omp_target_base_pointers_restrict_p): Delete.
13046 (scan_omp_target): Update call to scan_sharing_clauses.
13047 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
13049 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
13050 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
13051 (convert_local_omp_clauses): Likewise.
13052 * tree-pretty-print.c (dump_omp_clause): Likewise.
13053 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
13055 (omp_clause_code_name): Likewise.
13057 2018-06-20 Jakub Jelinek <jakub@redhat.com>
13060 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
13063 PR tree-optimization/86231
13064 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
13065 anti-range don't overwrite *vr0min before using it to compute *vr0max.
13067 2018-06-20 Tom de Vries <tdevries@suse.de>
13069 PR tree-optimization/86097
13070 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
13071 iv type if signedness of iv type is not the same as that of *nit.
13073 2018-06-20 Jakub Jelinek <jakub@redhat.com>
13075 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
13076 EDGE_EH edges, verify they are all EDGE_EH.
13078 2018-06-20 Maya Rashish <coypu@sdf.org>
13080 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
13082 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13084 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
13085 * config/aarch64/aarch64.c (xgene1_tunings): Add
13086 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
13087 (aarch64_mode_valid_for_sched_fusion_p):
13088 Allow 16-byte modes.
13089 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
13090 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
13092 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
13094 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
13095 * config/aarch64/iterators.md (VQ2): New mode iterator.
13097 2018-06-20 Martin Liska <mliska@suse.cz>
13099 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
13100 Change default ratio from 10 to 8.
13102 2018-06-20 Martin Liska <mliska@suse.cz>
13104 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
13106 (bit_test_cluster::find_bit_tests): Likewise.
13107 (switch_decision_tree::analyze_switch_statement): Find clusters.
13108 * tree-switch-conversion.h (struct jump_table_cluster): Document
13111 2018-06-20 Martin Liska <mliska@suse.cz>
13113 * tree-switch-conversion.c (switch_conversion::collect):
13114 Record m_uniq property.
13115 (switch_conversion::expand): Bail out for special conditions.
13116 (group_cluster::~group_cluster): New.
13117 (group_cluster::group_cluster): Likewise.
13118 (group_cluster::dump): Likewise.
13119 (jump_table_cluster::emit): New.
13120 (switch_decision_tree::fix_phi_operands_for_edges): New.
13121 (struct case_node): Remove struct.
13122 (jump_table_cluster::can_be_handled): New.
13123 (case_values_threshold): Moved to header.
13124 (reset_out_edges_aux): Likewise.
13125 (jump_table_cluster::is_beneficial): New.
13126 (bit_test_cluster::can_be_handled): Likewise.
13127 (add_case_node): Remove.
13128 (bit_test_cluster::is_beneficial): New.
13129 (case_bit_test::cmp): New.
13130 (bit_test_cluster::emit): New.
13131 (expand_switch_as_decision_tree_p): Remove.
13132 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
13133 (fix_phi_operands_for_edge): Likewise.
13134 (switch_decision_tree::analyze_switch_statement): New.
13135 (compute_cases_per_edge): Move ...
13136 (switch_decision_tree::compute_cases_per_edge): ... here.
13137 (try_switch_expansion): Likewise.
13138 (switch_decision_tree::try_switch_expansion): Likewise.
13139 (record_phi_operand_mapping): Likewise.
13140 (switch_decision_tree::record_phi_operand_mapping): Likewise.
13141 (emit_case_decision_tree): Likewise.
13142 (switch_decision_tree::emit): Likewise.
13143 (balance_case_nodes): Likewise.
13144 (switch_decision_tree::balance_case_nodes): Likewise.
13145 (dump_case_nodes): Likewise.
13146 (switch_decision_tree::dump_case_nodes): Likewise.
13147 (emit_jump): Likewise.
13148 (switch_decision_tree::emit_jump): Likewise.
13149 (emit_cmp_and_jump_insns): Likewise.
13150 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
13151 (emit_case_nodes): Likewise.
13152 (switch_decision_tree::emit_case_nodes): Likewise.
13153 (conditional_probability): Remove.
13154 * tree-switch-conversion.h (enum cluster_type): New.
13156 (struct cluster): Likewise.
13157 (cluster::cluster): Likewise.
13158 (struct simple_cluster): Likewise.
13159 (simple_cluster::simple_cluster): Likewise.
13160 (struct group_cluster): Likewise.
13161 (struct jump_table_cluster): Likewise.
13162 (struct bit_test_cluster): Likewise.
13163 (struct min_cluster_item): Likewise.
13164 (struct case_tree_node): Likewise.
13165 (case_tree_node::case_tree_node): Likewise.
13166 (jump_table_cluster::case_values_threshold): Likewise.
13167 (struct case_bit_test): Likewise.
13168 (struct switch_decision_tree): Likewise.
13169 (struct switch_conversion): Likewise.
13170 (switch_decision_tree::reset_out_edges_aux): Likewise.
13172 2018-06-20 Martin Liska <mliska@suse.cz>
13174 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
13175 (hoist_edge_and_branch_if_true): Likewise.
13176 (expand_switch_using_bit_tests_p): Likewise.
13177 (struct case_bit_test): Likewise.
13178 (case_bit_test_cmp): Likewise.
13179 (emit_case_bit_tests): Likewise.
13180 (switch_conversion::switch_conversion): New class.
13181 (struct switch_conv_info): Remove old struct.
13182 (collect_switch_conv_info): More to ...
13183 (switch_conversion::collect): ... this.
13184 (check_range): Likewise.
13185 (switch_conversion::check_range): Likewise.
13186 (check_all_empty_except_final): Likewise.
13187 (switch_conversion::check_all_empty_except_final): Likewise.
13188 (check_final_bb): Likewise.
13189 (switch_conversion::check_final_bb): Likewise.
13190 (create_temp_arrays): Likewise.
13191 (switch_conversion::create_temp_arrays): Likewise.
13192 (free_temp_arrays): Likewise.
13193 (gather_default_values): Likewise.
13194 (switch_conversion::gather_default_values): Likewise.
13195 (build_constructors): Likewise.
13196 (switch_conversion::build_constructors): Likewise.
13197 (constructor_contains_same_values_p): Likewise.
13198 (switch_conversion::contains_same_values_p): Likewise.
13199 (array_value_type): Likewise.
13200 (switch_conversion::array_value_type): Likewise.
13201 (build_one_array): Likewise.
13202 (switch_conversion::build_one_array): Likewise.
13203 (build_arrays): Likewise.
13204 (switch_conversion::build_arrays): Likewise.
13205 (gen_def_assigns): Likewise.
13206 (switch_conversion::gen_def_assigns): Likewise.
13207 (prune_bbs): Likewise.
13208 (switch_conversion::prune_bbs): Likewise.
13209 (fix_phi_nodes): Likewise.
13210 (switch_conversion::fix_phi_nodes): Likewise.
13211 (gen_inbound_check): Likewise.
13212 (switch_conversion::gen_inbound_check): Likewise.
13213 (process_switch): Use the newly created class.
13214 (switch_conversion::expand): New.
13215 (switch_conversion::~switch_conversion): New.
13216 * tree-switch-conversion.h: New file.
13218 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
13220 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
13221 tree-vect-patterns.c.
13222 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
13223 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
13224 (vect_recog_sad_pattern): Likewise.
13225 (vect_recog_widen_sum_pattern): Likewise.
13226 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
13227 (vect_recog_widen_shift_pattern): Remove the type_in argument.
13228 (vect_recog_rotate_pattern): Likewise.
13229 (vect_recog_mult_pattern): Likewise.
13230 (vect_recog_vector_vector_shift_pattern): Likewise.
13231 (vect_recog_divmod_pattern): Likewise.
13232 (vect_recog_mixed_size_cond_pattern): Likewise.
13233 (vect_recog_bool_pattern): Likewise.
13234 (vect_recog_mask_conversion_pattern): Likewise.
13235 (vect_try_gather_scatter_pattern): Likewise.
13236 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
13237 (vect_recog_over_widening_pattern): Likewise.
13238 (vect_recog_gather_scatter_pattern): Likewise.
13239 (vect_recog_func_ptr): Move from tree-vectorizer.h
13240 (vect_vect_recog_func_ptrs): Move further down the file.
13241 (vect_recog_func): Likewise. Remove the third argument.
13242 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
13243 (vect_pattern_recog_1): Expect the pattern function to do any
13244 necessary target tests. Also expect it to provide a vector type.
13245 Remove the type_in handling.
13247 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
13249 * tree-vect-patterns.c (vect_pattern_detected): New function.
13250 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
13251 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
13252 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
13253 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
13254 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
13255 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
13256 (vect_recog_mask_conversion_pattern)
13257 (vect_try_gather_scatter_pattern): Likewise.
13259 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
13261 * tree-vect-patterns.c (vect_get_internal_def): New function.
13262 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
13263 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
13264 (search_type_for_mask_1): Use it.
13266 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
13268 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
13269 redundant WIDEN_SUM_EXPR handling.
13270 (vect_recog_sad_pattern): Likewise.
13272 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
13274 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
13275 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
13276 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
13277 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
13278 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
13280 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
13282 * tree-vect-stmts.c (vectorizable_call): Make sure that we
13283 use the stmt_vec_info of the original bb statement for the
13284 new zero assignment, even if the call is part of a pattern.
13286 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
13288 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
13289 that the sequence is attached to the original statement rather
13290 than the pattern statement.
13291 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
13292 PATTERN_DEF_SEQ from the original statement rather than
13293 the main pattern statement.
13294 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
13295 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
13296 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
13298 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
13300 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
13301 definition statements before the early exit for statements that aren't
13303 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
13305 (vect_transform_loop): ...here. Process pattern definition
13306 statements without first checking whether the main pattern
13307 statement is live or relevant.
13309 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
13311 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
13312 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
13314 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
13316 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
13317 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
13318 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
13319 (expand_block_compare): Change select_block_compare_mode call.
13320 (expand_strncmp_align_check): Use new functions, fix comment.
13321 (emit_final_str_compare_gpr): New function.
13322 (expand_strn_compare): Refactor and clean up code.
13323 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
13325 2018-06-19 Tony Reix <tony.reix@atos.com>
13326 Damien Bergamini <damien.bergamini@atos.com>
13327 David Edelsohn <dje.gcc@gmail.com>
13329 * collect2.c (static_obj): New variable.
13330 (static_libs): New variable.
13331 (is_in_list): Uncomment declaration.
13332 (main): Track AIX libraries linked statically.
13333 (is_in_list): Uncomment definition.
13334 (scan_prog_file): Don't add AIX shared libraries initializer
13335 to constructor list if linking statically.
13337 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
13339 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
13341 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
13343 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
13345 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
13348 2018-06-19 Martin Liska <mliska@suse.cz>
13350 * config/i386/i386.c (ix86_can_inline_p): Do not use
13351 ipa_fn_summaries::get_create.
13352 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
13354 (devirtualization_time_bonus): Likewise.
13355 (ipcp_propagate_stage): Likewise.
13356 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
13357 (edge_set_predicate): Likewise.
13358 (evaluate_conditions_for_known_args): Likewise.
13359 (evaluate_properties_for_edge): Likewise.
13360 (ipa_call_summary::reset): Tranform to ...
13361 (ipa_call_summary::~ipa_call_summary): ... this.
13362 (ipa_fn_summary::reset): Transform to ...
13363 (ipa_fn_summary::~ipa_fn_summary): ... this.
13364 (ipa_fn_summary_t::remove): Rename to ...
13365 (ipa_fn_summary_t::remove_callees): ... this.
13366 (ipa_fn_summary_t::duplicate): Use placement new
13367 instead of memory copy.
13368 (ipa_call_summary_t::duplicate): Likewise.
13369 (ipa_call_summary_t::remove): Remove.
13370 (dump_ipa_call_summary): Change get_create to get.
13371 (ipa_dump_fn_summary): Dump only when summary exists.
13372 (analyze_function_body): Use symbol_summary::get instead
13374 (compute_fn_summary): Likewise.
13375 (estimate_edge_devirt_benefit): Likewise.
13376 (estimate_edge_size_and_time): Likewise.
13377 (inline_update_callee_summaries): Likewise.
13378 (remap_edge_change_prob): Likewise.
13379 (remap_edge_summaries): Likewise.
13380 (ipa_merge_fn_summary_after_inlining): Likewise.
13381 (write_ipa_call_summary): Likewise.
13382 (ipa_fn_summary_write): Likewise.
13383 (ipa_free_fn_summary): Likewise.
13384 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
13385 (struct ipa_call_summary): Likewise.
13386 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
13388 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
13389 (estimate_size_after_inlining): Likewise.
13390 (estimate_growth): Likewise.
13391 (growth_likely_positive): Likewise.
13392 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
13393 (inline_call): Likewise.
13394 * ipa-inline.c (caller_growth_limits): Likewise.
13395 (can_inline_edge_p): Likewise.
13396 (can_inline_edge_by_limits_p): Likewise.
13397 (compute_uninlined_call_time): Likewise.
13398 (compute_inlined_call_time): Likewise.
13399 (want_inline_small_function_p): Likewise.
13400 (edge_badness): Likewise.
13401 (update_caller_keys): Likewise.
13402 (update_callee_keys): Likewise.
13403 (inline_small_functions): Likewise.
13404 (inline_to_all_callers_1): Likewise.
13405 (dump_overall_stats): Likewise.
13406 (early_inline_small_functions): Likewise.
13407 (early_inliner): Likewise.
13408 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
13409 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
13410 * ipa-pure-const.c (malloc_candidate_p): Likewise.
13411 * ipa-split.c (execute_split_functions): Likewise.
13412 * symbol-summary.h: Likewise.
13413 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
13415 2018-06-19 Richard Biener <rguenther@suse.de>
13417 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
13418 (vectorize_loops): ... here. Fix dbgcnt handling.
13419 (try_vectorize_loop): Wrap try_vectorize_loop_1.
13421 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
13424 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
13425 ieee128 argument takes up only one (vector) register, not two (floating
13428 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
13430 * gimplify.c (gimplify_init_constructor): Really never clear for an
13431 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
13433 2018-06-19 Richard Biener <rguenther@suse.de>
13435 PR tree-optimization/86179
13436 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
13437 after failed recognition.
13439 2018-06-18 Martin Sebor <msebor@redhat.com>
13441 PR middle-end/85602
13442 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
13443 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
13444 Handle integer subtraction.
13445 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
13446 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
13448 2018-06-18 David Malcolm <dmalcolm@redhat.com>
13450 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
13451 param from rtx to rtx_insn *.
13452 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
13454 (frv_ifcvt_modify_insn): Likwise.
13455 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
13456 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
13457 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
13458 as_a <rtx_insn *> cast to local "unprotected_region" once
13459 it's been established that it's not NULL or pc_rtx.
13460 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
13461 param "sethi" from rtx to rtx_insn *.
13462 (nds32_group_float_insns): Likewise for param "insn".
13463 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
13465 (vax_output_int_subtract): Likewise.
13466 * config/vax/vax.c (vax_output_int_add): Likewise for param
13468 (vax_output_int_subtract): Likewise.
13469 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
13470 (emit_pattern_after): Likewise for param "after".
13471 (emit_insn_after): Likewise.
13472 (emit_jump_insn_after): Likewise.
13473 (emit_call_insn_after): Likewise.
13474 (emit_debug_insn_after): Likewise.
13475 (emit_pattern_before): Likewise for param "before".
13476 (emit_insn_before): Likewise.
13477 (emit_jump_insn_before): Likewise.
13478 * final.c (get_insn_template): Likewise for param "insn", removing
13480 * output.h (get_insn_template): Likewise for 2nd param.
13481 * rtl.h (emit_insn_before): Likewise.
13482 (emit_jump_insn_before): Likewise.
13483 (emit_debug_insn_before_noloc): Likewise.
13484 (emit_insn_after): Likewise.
13485 (emit_jump_insn_after): Likewise.
13486 (emit_call_insn_after): Likewise.
13487 (emit_debug_insn_after): Likewise.
13488 (set_insn_deleted): Likewise for param.
13490 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
13493 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
13494 floating point modes, so that IFmode is numerically greater than
13495 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
13496 to declare the ordering. This prevents IFmode from being
13497 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
13498 machine. Include rs6000-modes.h to share the fractional values
13499 between genmodes* and the rest of the compiler.
13500 (IFmode): Likewise.
13501 (KFmode): Likewise.
13502 (TFmode): Likewise.
13503 * config/rs6000/rs6000-modes.h: New file.
13504 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
13505 meaning of rs6000_long_double_size so that 126..128 selects an
13506 appropriate 128-bit floating point type.
13507 (rs6000_option_override_internal): Likewise.
13508 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
13509 (TARGET_LONG_DOUBLE_128): Change the meaning of
13510 rs6000_long_double_size so that 126..128 selects an appropriate
13511 128-bit floating point type.
13512 (LONG_DOUBLE_TYPE_SIZE): Update comment.
13513 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
13514 source and destination to match the standard usage.
13515 (truncifkf2): Likewise.
13516 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
13517 ISA 2.07 to use an explicit clobber, instead of passing in a
13519 (copysign<mode>3_soft): Likewise.
13521 2018-06-18 David Malcolm <dmalcolm@redhat.com>
13523 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
13524 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
13525 (vect_slp_analyze_instance_dependence): Likewise.
13526 (vect_enhance_data_refs_alignment): Likewise.
13527 (vect_analyze_data_refs_alignment): Likewise.
13528 (vect_slp_analyze_and_verify_instance_alignment
13529 (vect_analyze_data_ref_accesses): Likewise.
13530 (vect_prune_runtime_alias_test_list): Likewise.
13531 (vect_analyze_data_refs): Likewise.
13532 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
13533 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
13534 (vect_analyze_scalar_cycles_1): Likewise.
13535 (vect_get_loop_niters): Likewise.
13536 (vect_analyze_loop_form_1): Likewise.
13537 (vect_update_vf_for_slp): Likewise.
13538 (vect_analyze_loop_operations): Likewise.
13539 (vect_analyze_loop): Likewise.
13540 (vectorizable_induction): Likewise.
13541 (vect_transform_loop): Likewise.
13542 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
13543 * tree-vect-slp.c (vect_analyze_slp): Likewise.
13544 (vect_make_slp_decision): Likewise.
13545 (vect_detect_hybrid_slp): Likewise.
13546 (vect_slp_analyze_operations): Likewise.
13547 (vect_slp_bb): Likewise.
13548 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
13549 (vectorizable_bswap): Likewise.
13550 (vectorizable_call): Likewise.
13551 (vectorizable_simd_clone_call): Likewise.
13552 (vectorizable_conversion): Likewise.
13553 (vectorizable_assignment): Likewise.
13554 (vectorizable_shift): Likewise.
13555 (vectorizable_operation): Likewise.
13556 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
13558 2018-06-18 Martin Sebor <msebor@redhat.com>
13560 PR tree-optimization/81384
13561 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
13562 * builtins.c (expand_builtin_strnlen): New function.
13563 (expand_builtin): Call it.
13564 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
13565 * builtins.def (BUILT_IN_STRNLEN): New.
13566 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
13567 Warn for bounds in excess of maximum object size.
13568 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
13569 single-value ranges. Handle strnlen.
13570 (handle_builtin_strlen): Handle strnlen.
13571 (strlen_check_and_optimize_stmt): Same.
13572 * doc/extend.texi (Other Builtins): Document strnlen.
13574 2018-06-18 Maya Rashish <coypu@sdf.org>
13576 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
13577 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
13578 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
13580 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
13582 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
13584 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13586 * tree.c (escaped_string::escape): Replace cast to char * by
13587 const_cast<char *> (unescaped).
13589 2018-06-18 Nick Clifton <nickc@redhat.com>
13592 * tree.c (escaped_string): New class. Converts an unescaped
13593 string into its escaped equivalent.
13594 (warn_deprecated_use): Use the new class to convert the
13595 deprecation message, if present.
13596 (test_escaped_strings): New self test.
13597 (test_c_tests): Add test_escaped_strings.
13598 * doc/extend.texi (deprecated): Add a note that the
13599 deprecation message is affected by the -fmessage-length
13600 option, and that control characters will be escaped.
13601 (#pragma GCC error): Document this pragma.
13602 (#pragma GCC warning): Likewise.
13603 * doc/invoke.texi (-fmessage-length): Document this option's
13604 effect on the #warning and #error preprocessor directives and
13605 the deprecated attribute.
13607 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
13609 * tree.c (decl_value_expr_lookup): Revert latest change.
13610 (decl_value_expr_insert): Likewise.
13612 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
13614 * gimplify.c (nonlocal_vlas): Delete.
13615 (nonlocal_vla_vars): Likewise.
13616 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
13618 (gimplify_body): Do not create and destroy nonlocal_vlas.
13619 * tree-nested.c: Include diagnostic.h.
13620 (use_pointer_in_frame): Tweak.
13621 (lookup_field_for_decl): Add assertion and declare the transformation.
13622 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
13623 internal error when the reference is in a wrong context. Do not
13624 create a debug decl by default.
13625 (note_nonlocal_block_vlas): Delete.
13626 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
13627 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
13628 create a debug decl by default.
13629 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
13630 call is in a wrong context.
13631 (fixup_vla_decls): New function.
13632 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
13633 debug variables were created.
13634 * tree.c (decl_value_expr_lookup): Add checking assertion.
13635 (decl_value_expr_insert): Likewise.
13637 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
13639 PR middle-end/82479
13640 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
13641 * tree-scalar-evolution.c (interpret_expr): Likewise.
13642 (expression_expensive_p): Likewise.
13643 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
13644 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
13645 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
13646 (ssa_defined_by_minus_one_stmt_p): New.
13648 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
13650 PR middle-end/64946
13651 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
13652 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
13653 * dojump.c (do_jump): Likewise.
13654 * expr.c (expand_expr_real_2): Check operand type's sign.
13655 * fold-const.c (const_unop): Handle ABSU_EXPR.
13656 (fold_abs_const): Likewise.
13657 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
13658 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
13659 (strip_sign_op_1): Likesise.
13660 * match.pd: Add new pattern to generate ABSU_EXPR.
13661 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
13662 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
13663 * tree-eh.c (operation_could_trap_helper_p): Likewise.
13664 * tree-inline.c (estimate_operator_cost): Likewise.
13665 * tree-pretty-print.c (dump_generic_node): Likewise.
13666 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
13667 * tree.def (ABSU_EXPR): New.
13669 2018-06-16 Jakub Jelinek <jakub@redhat.com>
13671 PR middle-end/86095
13672 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
13673 documented as preserved for backward compatibility only.
13674 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
13676 PR rtl-optimization/86108
13677 * bb-reorder.c (create_forwarder_block): Renamed to ...
13678 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
13679 jump from new landing pad to the second part.
13680 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
13683 2018-06-15 Jakub Jelinek <jakub@redhat.com>
13685 PR middle-end/85878
13686 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
13687 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
13688 Only call store_expr for halves if the mode is the same.
13690 PR middle-end/86123
13691 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
13692 Fix up comment formatting.
13694 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
13696 * typed-splay-tree.h (typed_splay_tree::remove): New function.
13697 (typed_splay_tree::closure,
13698 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
13699 (typed_splay_tree::typed_splay_tree,
13700 typed_splay_tree::operator =): Declared private.
13701 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
13702 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
13703 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
13704 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
13705 typed_splay_tree::splay_tree_splay,
13706 typed_splay_tree::splay_tree_foreach_helper,
13707 typed_splay_tree::splay_tree_insert,
13708 typed_splay_tree::splay_tree_remove,
13709 typed_splay_tree::splay_tree_lookup,
13710 typed_splay_tree::splay_tree_predecessor,
13711 typed_splay_tree::splay_tree_successor,
13712 typed_splay_tree::splay_tree_min,
13713 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
13714 (typed_splay_tree::root, typed_splay_tree::comp,
13715 typed_splay_tree::delete_key,
13716 typed_splay_tree::delete_value): New data members.
13717 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
13718 typed_splay_tree::remove.
13720 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
13722 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
13723 -mginv and -mno-ginv to the assembler.
13724 * config/mips/mips.opt (-mcrc): New option.
13725 (-mginv): Likewise.
13726 * doc/invoke.text (-mcrc): Document.
13727 (-mginv): Likewise.
13729 2018-06-15 Nick Clifton <nickc@redhat.com>
13732 * tree.c (escaped_string): New class. Converts an unescaped
13733 string into its escaped equivalent.
13734 (warn_deprecated_use): Use the new class to convert the
13735 deprecation message, if present.
13736 (test_escaped_strings): New self test.
13737 (test_c_tests): Add test_escaped_strings.
13738 * doc/extend.texi (deprecated): Add a note that the
13739 deprecation message is affected by the -fmessage-length
13740 option, and that control characters will be escaped.
13741 (#pragma GCC error): Document this pragma.
13742 (#pragma GCC warning): Likewise.
13743 * doc/invoke.texi (-fmessage-length): Document this option's
13744 effect on the #warning and #error preprocessor directives and
13745 the deprecated attribute.
13747 2018-06-15 Richard Biener <rguenther@suse.de>
13749 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
13750 here, also noting vector size used.
13751 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
13752 size used in MSG_OPTIMIZED_LOCATIONS dump.
13753 (pass_slp_vectorize::execute): Adjust.
13755 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
13758 * config/arc/arc.c (arc_return_address_register): Fix
13761 2018-06-15 Richard Biener <rguenther@suse.de>
13763 PR middle-end/86159
13764 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
13765 leave useless conversion stripping to force_gimple_operand_gsi.
13766 (gimplify_build2): Likewise.
13767 (gimplify_build1): Likewise.
13769 2018-06-15 Richard Biener <rguenther@suse.de>
13771 PR middle-end/86076
13772 * tree-cfg.c (move_stmt_op): unshare invariant addresses
13773 before adjusting their block.
13775 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
13777 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
13778 multilibs for *-*-rtems*.
13779 * config/riscv/t-rtems: New file.
13781 2018-06-14 Jakub Jelinek <jakub@redhat.com>
13783 PR middle-end/86122
13784 * match.pd ((A +- CST1) +- CST2): Punt if last resort
13785 unsigned_type_for returns NULL.
13788 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
13789 subregs of multi-word pseudos unless the float mode has word size.
13791 2018-06-14 Richard Biener <rguenther@suse.de>
13793 PR middle-end/86139
13794 * tree-vect-generic.c (build_word_mode_vector_type): Remove
13795 duplicate and harmful type_hash_canon.
13796 * tree.c (type_hash_canon): Assert we didn't find ourselves.
13798 2018-06-14 Richard Biener <rguenther@suse.de>
13801 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
13804 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
13806 * config/rtems.h (STDINT_LONG32): Define.
13808 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
13809 Prachi Godbole <prachi.godbole@imgtec.com>
13811 * config/mips/mips-cpus.def: Define P6600.
13812 * config/mips/mips-tables.opt: Regenerate.
13813 * config/mips/mips.c (mips_ucbranch_type): New enum.
13814 (mips_rtx_cost_data): Add support for P6600.
13815 (mips_issue_rate): Likewise.
13816 (mips_multipass_dfa_lookahead): Likewise.
13817 (mips_avoid_hazard): Likewise.
13818 (mips_reorg_process_insns): Likewise.
13819 (mips_classify_branch_p6600): New function.
13820 * config/mips/mips.h (TUNE_P6600): New define.
13821 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
13822 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
13823 * config/mips/mips.md: Include p6600.md.
13824 (processor): Add p6600.
13825 * config/mips/p6600.md: New file.
13826 * doc/invoke.texi: Add p6600 to supported architectures.
13828 2018-06-13 Martin Sebor <msebor@redhat.com>
13830 PR tree-optimization/86114
13831 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
13833 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
13835 2018-06-13 Richard Biener <rguenther@suse.de>
13837 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
13838 Properly set vector type of the intermediate stmt.
13839 * tree-vect-stmts.c (vectorizable_operation): The destination
13840 var always has vectype_out type.
13842 2018-06-13 Jeff Law <law@redhat.com>
13844 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
13845 integer 0 for argument to print_rtl_with_bb.
13846 (rl78_reorg): Likewise.
13848 2018-06-13 David Malcolm <dmalcolm@redhat.com>
13850 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
13851 from rtx to rtx_insn *.
13852 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
13854 (add_sched_insns_for_speculation): Likewise for local "target",
13855 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
13856 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
13857 from rtx_insn ** to rtx_code_label **.
13858 (reorg_emit_nops): Likewise.
13859 (c6x_reorg): Likewise for local "call_labels".
13860 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
13862 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
13863 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
13864 the loops over LABEL_REFS.
13865 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
13867 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
13868 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
13869 (split_branches): Strengthen local "olabel" from rtx to
13870 rtx_insn *, adding a safe_as_a cast.
13871 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
13873 (add_insn_after): Likewise for first two params.
13874 (add_insn_before): Likewise.
13875 (remove_insn): Likewise for param.
13876 (emit_pattern_before_noloc): Likewise for second and third params.
13877 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
13878 (emit_call_insn_before_noloc): Likewise.
13879 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
13881 (emit_barrier_before): Likewise.
13882 (emit_label_before): Strengthen "label" param from "rtx" to
13883 "rtx_code_label *". Strengthen "before" param from "rtx" to
13885 (emit_insn_after_1): Strengthen "after" param from "rtx" to
13887 (emit_pattern_after_noloc): Likewise.
13888 (emit_insn_after_noloc): Likewise.
13889 (emit_jump_insn_after_noloc): Likewise.
13890 (emit_call_insn_after_noloc): Likewise.
13891 (emit_debug_insn_after_noloc): Likewise.
13892 (emit_barrier_after): Likewise.
13893 (emit_label_after): Likewise for both params.
13894 (emit_pattern_after_setloc): Likewise for "after" param. Convert
13895 "loc" param from "int" to "location_t".
13896 (emit_insn_after_setloc): Likewise.
13897 (emit_jump_insn_after_setloc): Likewise.
13898 (emit_call_insn_after_setloc): Likewise.
13899 (emit_debug_insn_after_setloc): Likewise.
13900 (emit_pattern_before_setloc): Likewise for "before" param. Convert
13901 "loc" param from "int" to "location_t".
13902 (emit_pattern_before): Convert NULL_RTX to NULL.
13903 (emit_insn_before_setloc): Convert "loc" param from "int" to
13905 (emit_jump_insn_before_setloc): Likewise.
13906 (emit_call_insn_before_setloc): Likewise.
13907 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
13908 rtx_insn *. Convert "loc" param from "int" to "location_t".
13909 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
13910 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
13911 Convert 3rd param from "int" to "location_t".
13912 (emit_barrier_before, emit_barrier_after, next_real_insn):
13913 Strengthen param from rtx to rtx_insn *.
13914 (emit_label_before): Strengthen 1st param from "rtx" to
13915 "rtx_code_label *". Strengthen 2nd param from "rtx" to
13917 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
13918 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
13919 Strengthen 2nd param from "rtx" to "rtx_insn *".
13920 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
13921 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
13922 Likewise. Convert 3rd param from "int" to "location_t".
13923 (emit_label_after): Strengthen 1st param from "rtx" to
13924 "rtx_code_label *".
13925 (next_real_insn, remove_insn): Strengthen param from "rtx" to
13927 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
13928 from "rtx" to "rtx_insn *".
13930 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
13932 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
13933 bodies streamed in with -Q.
13934 * dumpfile.c (dump_files): Add lto-stream-out dump file.
13935 * dumpfile.h (tree_dump_index): Add lto_stream_out.
13936 * gimple-streamer-out.c: Include gimple-pretty-print.h
13937 (output_bb): Dump stmts streamed.
13938 * lto-section-out.c: Include print-tree.h
13939 (lto_begin_section): Dump sections created.
13940 (lto_output_decl_index): Dump decl encoded.
13941 * lto-streamer-out.c: Include print-tree.h
13942 (create_output_block): Dump output block created.
13943 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
13944 (output_function): Dump function output.
13945 (output_constructor): Dump constructor streamed.
13946 (write_global_stream): Output indexes encoded.
13947 (produce_asm_for_decls): Dump streams encoded.
13948 * lto-streamer.c (streamer_dump_file): New global var.
13949 * lto-streamer.h (streamer_dump_file): Declare.
13950 * passes.c (ipa_write_summaries): Initialize streamer dump.
13951 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
13954 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
13957 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
13958 offsets for register save directives. Emit a second batch of save
13959 directives, if need be, when the function accesses prior frames.
13961 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13963 * config/arc/fpu.md (fmasf4): Force operand to register.
13964 (fnmasf4): Likewise.
13966 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13968 * config/arc/arc-protos.h (arc_pad_return): Remove.
13969 * config/arc/arc.c (machine_function): Remove force_short_suffix
13971 (arc_print_operand): Adjust printing of '&'.
13972 (arc_verify_short): Remove conditional printing of short suffix.
13973 (arc_final_prescan_insn): Remove reference to size_reason.
13974 (pad_return): New function.
13975 (arc_reorg): Call pad_return.
13976 (arc_pad_return): Remove.
13977 (arc_init_machine_status): Remove reference to force_short_suffix.
13978 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
13979 (attr length): When attribute iscompact is true force to 2
13980 regardless; in the case of maybe check if we want to force the
13981 instruction to have 4 bytes length.
13982 (nopv): Change it to generate 4 byte long nop as well.
13983 (blockage): New pattern.
13984 (simple_return): Remove call to arc_pad_return.
13985 (p_return_i): Likewise.
13987 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13989 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
13991 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13993 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
13996 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13998 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
13999 for ARC700 and ARCv2.
14001 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
14004 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
14005 operands[2] instead of operands[1].
14008 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
14010 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
14011 case, check whether the outer register overlaps an unallocatable
14012 register, not just whether it fits the required class.
14014 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
14016 * poly-int.h (can_div_trunc_p): Add new overload in which all values
14018 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
14019 (memrefs_conflict_p): Likewise.
14020 (init_alias_analysis): Likewise.
14021 * cfgexpand.c (expand_debug_expr): Likewise.
14022 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
14023 * cse.c (fold_rtx): Likewise.
14024 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
14025 * expr.c (emit_block_move_hints): Likewise.
14026 (clear_storage_hints, push_block, emit_push_insn): Likewise.
14027 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
14028 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
14029 (emit_group_store): Likewise.
14030 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
14031 to read the PRE/POST_MODIFY increment.
14032 * calls.c (store_one_arg): Use strip_offset.
14033 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
14035 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
14037 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
14038 (simplify_binary_operation_1): Extend CONST_INT handling to
14040 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
14041 than a HOST_WIDE_INT.
14042 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
14044 (adjust_mems, add_stores): Update accodingly.
14045 (vt_canonicalize_addr): Track polynomial offsets.
14046 (emit_note_insn_var_location): Likewise.
14047 (vt_add_function_parameter): Likewise.
14048 (vt_initialize): Likewise.
14050 2018-06-12 Jeff Law <law@redhat.com>
14052 * config.gcc (alpha*-*-freebsd*): Remove.
14053 * config/alpha/freebsd.h: Remove.
14055 2018-06-12 David Malcolm <dmalcolm@redhat.com>
14058 * spellcheck-tree.c (levenshtein_distance): Rename to...
14059 (get_edit_distance): ...this, and update for underlying renaming.
14060 * spellcheck-tree.h (levenshtein_distance): Rename to...
14061 (get_edit_distance): ...this.
14062 * spellcheck.c (levenshtein_distance): Rename to...
14063 (get_edit_distance): ...this. Convert from Levenshtein distance
14064 to Damerau-Levenshtein distance by supporting transpositions of
14065 adjacent characters. Rename "v1" to "v_next" and "v0" to
14067 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
14068 (selftest::test_edit_distance_unit_test_oneway): ...this, and
14069 update for underlying renaming.
14070 (selftest::levenshtein_distance_unit_test): Rename to...
14071 (selftest::test_get_edit_distance_unit): ...this, and update for
14072 underlying renaming.
14073 (selftest::test_find_closest_string): Add example from PR 69968
14074 where transposition helps
14075 (selftest::test_metric_conditions): Update for renaming.
14076 (selftest::test_metric_conditions): Likewise.
14077 (selftest::spellcheck_c_tests): Likewise.
14078 * spellcheck.h (levenshtein_distance): Rename both overloads to...
14079 (get_edit_distance): ...this.
14080 (best_match::consider): Update for renaming.
14082 2018-06-12 Martin Sebor <msebor@redhat.com>
14084 PR tree-optimization/85259
14085 * builtins.c (compute_objsize): Handle constant offsets.
14086 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
14087 true iff a warning has been issued.
14088 * gimple.h (gimple_nonartificial_location): New function.
14089 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
14090 gimple_nonartificial_location and handle -Wno-system-headers.
14091 (handle_builtin_stxncpy): Same.
14093 2018-06-12 Martin Sebor <msebor@redhat.com>
14096 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
14098 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14100 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14101 BUILTIN_VEC_XST entries for pointer to double and long long.
14103 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
14106 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
14108 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
14111 2018-06-12 Martin Liska <mliska@suse.cz>
14113 * doc/options.texi: Document IntegerRange.
14115 2018-06-12 Martin Liska <mliska@suse.cz>
14117 * config/i386/i386.opt: Make MPX-related options as Deprecated.
14118 * opt-functions.awk: Handle Deprecated flag.
14119 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
14121 (read_cmdline_option): Report warning for a deprecated option.
14122 * opts.h (struct cl_option): Add new field cl_deprecated.
14123 (CL_ERR_DEPRECATED): New.
14125 2018-06-12 Martin Liska <mliska@suse.cz>
14127 * doc/options.texi: Document Deprecated option flag.
14129 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
14131 * config/arc/arc-arch.h (arc_extras): New enum.
14132 (arc_cpu_t):Add field extra.
14133 (arc_cpu_types): Consider the extras.
14134 * config/arc/arc-cpus.def: Add extras info.
14135 * config/arc/arc-opts.h (processor_type): Consider extra field.
14136 * config/arc/arc.c (arc_override_options): Handle extra field.
14138 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
14140 * config/arc/arc-arch.h: Update ARC_OPTX macro.
14141 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
14143 * config/arc/arc.c (arc_init): Update pic warning.
14144 (irq_range): Update irq range parsing warnings.
14145 (arc_override_options): Update various warning messages.
14146 (arc_handle_aux_attribute): Likewise.
14148 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
14150 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
14152 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
14154 * doc/sourcebuild.texi: Document usage of line number 0 in verify
14155 compiler messages directives.
14157 2018-06-12 Matthew Fortune <mfortune@gmail.com>
14159 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
14160 * config/mips/mips-tables.opt: Regenerate.
14161 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
14163 * doc/invoke.texi: Document -march=i6500.
14165 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
14167 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
14168 (i6400_gpmul): Add cpu_unit.
14169 (i6400_gpdiv): Likewise.
14170 (i6400_msa_add_d): Update reservations.
14171 (i6400_msa_int_add) Likewise.
14172 (i6400_msa_short_logic3) Likewise.
14173 (i6400_msa_short_logic2) Likewise.
14174 (i6400_msa_short_logic) Likewise.
14175 (i6400_msa_move) Likewise.
14176 (i6400_msa_cmp) Likewise.
14177 (i6400_msa_short_float2) Likewise.
14178 (i6400_msa_div_d) Likewise.
14179 (i6400_msa_long_logic1) Likewise.
14180 (i6400_msa_long_logic2) Likewise.
14181 (i6400_msa_mult) Likewise.
14182 (i6400_msa_long_float2) Likewise.
14183 (i6400_msa_long_float4) Likewise.
14184 (i6400_msa_long_float5) Likewise.
14185 (i6400_msa_long_float8) Likewise.
14186 (i6400_fpu_fadd): Include frint type.
14187 (i6400_fpu_store): New define_insn_reservation.
14188 (i6400_fpu_load): Likewise.
14189 (i6400_fpu_move): Likewise.
14190 (i6400_fpu_fcmp): Likewise.
14191 (i6400_fpu_fmadd): Likewise.
14192 (i6400_int_mult): Include imul3nc type and update reservation.
14193 (i6400_int_div): Include idiv3 type and update reservation.
14194 (i6400_int_load): Update to check type not move_type.
14195 (i6400_int_store): Likewise.
14196 (i6400_int_prefetch): Set zero latency.
14198 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
14200 * gcc.c: Document new %@{...} sequence.
14201 (LINK_COMMAND_SPEC): Use it for the -L switches.
14202 (cpp_unique_options): Use it for the -I switches.
14203 (at_file_argbuf): New global variable.
14204 (in_at_file): Likewise.
14205 (alloc_args): Create at_file_argbuf.
14206 (clear_args): Truncate at_file_argbuf.
14207 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
14208 (open_at_file): New function.
14209 (close_at_file): Likewise.
14210 (create_at_file): Delete.
14211 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
14214 (validate_switches_from_spec): Deal with %@{...} sequence.
14215 (validate_switches): Likewise.
14216 (driver::finalize): Call clear_args.
14218 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
14220 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
14222 2018-06-11 Martin Sebor <msebor@redhat.com>
14224 * doc/invoke.texi (-Wall): List -Wc++17-compat.
14225 (Wno-class-memaccess): Add @opindex.
14226 (Wno-templates, Wno-multiple-inheritance): Same.
14227 (Wno-virtual-inheritance, Wno-namespaces): Same.
14228 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
14229 (Wno-format-overflow, Wno-format-truncation): Same.
14230 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
14231 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
14232 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
14233 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
14234 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
14235 (Wno-misspelled-isr): Same.
14237 2018-06-11 Martin Sebor <msebor@redhat.com>
14239 * PR tree-optimization/86083
14240 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
14242 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
14244 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
14246 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
14249 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
14250 on the correct operand.
14251 (*movdi_internal64): Ditto.
14253 2018-06-11 Martin Liska <mliska@suse.cz>
14255 PR tree-optimization/86089
14256 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
14258 2018-06-11 Julia Koval <julia.koval@intel.com>
14260 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
14261 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
14262 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
14264 2018-06-11 Olivier Hainque <hainque@adacore.com>
14266 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
14267 for Ada with strict dwarf2.
14269 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
14272 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
14275 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
14277 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
14279 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
14281 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
14284 2018-06-08 Martin Liska <mliska@suse.cz>
14286 * tree-cfg.h (debug_function): Fix argument type to match
14289 2018-06-08 Martin Liska <mliska@suse.cz>
14291 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
14292 Remove usage of MPX-related (and removed) fields.
14293 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
14295 2018-06-08 David Malcolm <dmalcolm@redhat.com>
14297 * cfg.c (debug): Use TDF_NONE rather than 0.
14298 * cfghooks.c (debug): Likewise.
14299 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
14300 (struct dump_option_value_info): Convert to...
14301 (struct kv_pair): ...this template type.
14302 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
14304 (optinfo_verbosity_options): Likewise.
14305 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
14307 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
14308 than int for "optgroup_flags" param.
14309 (dump_generic_expr_loc): Use dump_flags_t rather than int for
14311 (dump_dec): Likewise.
14312 (dump_finish): Use TDF_NONE rather than 0.
14313 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
14314 rather than int for "optgroup_flags" param. Use TDF_NONE rather
14315 than 0. Update for change to option_ptr.
14316 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
14317 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
14318 0. Update for changes to optinfo_verbosity_options and
14320 (opt_info_switch_p): Convert optgroup_flags from int to
14322 (dump_basic_block): Use dump_flags_t rather than int
14323 for "dump_kind" param.
14324 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
14325 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
14326 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
14327 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
14328 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
14329 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
14330 TDF_NONE): Convert from macros to...
14331 (enum dump_flag): ...this new enum.
14332 (dump_flags_t): Update to use enum.
14333 (operator|, operator&, operator~, operator|=, operator&=):
14334 Implement for dump_flags_t.
14335 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
14336 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
14337 Convert from macros to...
14338 (enum optgroup_flag): ...this new enum.
14339 (optgroup_flags_t): New typedef.
14340 (operator|, operator|=): Implement for optgroup_flags_t.
14341 (struct dump_file_info): Convert field "alt_flags" to
14342 dump_flags_t. Convert field "optgroup_flags" to
14344 (dump_basic_block): Use dump_flags_t rather than int for param.
14345 (dump_generic_expr_loc): Likewise.
14346 (dump_dec): Likewise.
14347 (dump_register): Convert param "optgroup_flags" to
14349 (opt_info_enable_passes): Likewise.
14350 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
14352 * gimple-pretty-print.c (debug): Likewise.
14353 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
14354 (merged_store_group::apply_stores): Likewise.
14355 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
14356 * gimple.c (verify_gimple_pp): Likewise.
14357 * graphite-poly.c (print_pbb_body): Likewise.
14358 * passes.c (pass_manager::register_one_dump_file): Convert
14359 local "optgroup_flags" to optgroup_flags_t.
14360 * print-tree.c (print_node): Use TDF_NONE rather than 0.
14362 (debug_body): Likewise.
14363 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
14364 to optgroup_flags_t.
14365 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
14367 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
14368 (convert_mult_to_fma): Likewise.
14369 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
14370 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
14371 * tree-vect-data-refs.c (dump_lower_bound): Convert param
14372 "dump_kind" to dump_flags_t.
14374 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
14376 * config/rs6000/rs6000.c (min, max): Delete.
14378 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
14380 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
14381 -mabi=spe and -mabi=no-spe.
14383 2018-06-08 Martin Liska <mliska@suse.cz>
14385 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
14386 where we expect an existing summary.
14388 2018-06-08 Martin Liska <mliska@suse.cz>
14390 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
14391 * ipa-inline.h (estimate_edge_growth): Likewise.
14393 2018-06-08 Martin Liska <mliska@suse.cz>
14395 * cgraph.c (function_version_hasher::hash): Use
14396 cgraph_node::get_uid ().
14397 (function_version_hasher::equal):
14398 * cgraph.h (cgraph_node::get_uid): New method.
14399 * ipa-inline.c (update_caller_keys): Use
14400 cgraph_node::get_uid ().
14401 (update_callee_keys): Likewise.
14402 * ipa-utils.c (searchc): Likewise.
14403 (ipa_reduced_postorder): Likewise.
14404 * lto-cgraph.c (input_node): Likewise.
14405 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
14406 * symbol-summary.h (symtab_insertion): Likewise.
14407 (symtab_removal): Likewise.
14408 (symtab_duplication): Likewise.
14409 * tree-pretty-print.c (dump_function_header): Likewise.
14410 * tree-sra.c (convert_callers_for_node): Likewise.
14412 2018-06-08 Martin Liska <mliska@suse.cz>
14414 * cgraph.c (symbol_table::create_edge): Always assign a new
14416 (symbol_table::free_edge): Do not recycle numbers.
14417 * cgraph.h (cgraph_edge::get): New method.
14418 * symbol-summary.h (symtab_removal): Use it.
14419 (symtab_duplication): Likewise.
14420 (call_summary::hashable_uid): Remove.
14422 2018-06-08 Martin Liska <mliska@suse.cz>
14424 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
14425 (initialize_growth_caches): Remove.
14426 (free_growth_caches): Likewise.
14427 (do_estimate_edge_time): Use edge_growth_cache.
14428 (do_estimate_edge_size): Likewise.
14429 (do_estimate_edge_hints): Likewise.
14430 * ipa-inline.c (reset_edge_caches): Likewise.
14431 (recursive_inlining): Likewise.
14432 (inline_small_functions): Likewise.
14433 * ipa-inline.h (initialize_growth_caches): Remove.
14434 (estimate_edge_size): Likewise.
14435 (estimate_edge_time): Likewise.
14436 (estimate_edge_hints): Likewise.
14437 (reset_edge_growth_cache): Likewise.
14438 * symbol-summary.h (call_summary::remove): New method.
14440 2018-06-08 Martin Liska <mliska@suse.cz>
14442 * ipa-cp.c (class edge_clone_summary): New summary.
14443 (grow_edge_clone_vectors): Remove.
14444 (ipcp_edge_duplication_hook): Remove.
14445 (class edge_clone_summary_t): New call_summary class.
14446 (ipcp_edge_removal_hook): Remove.
14447 (edge_clone_summary_t::duplicate): New function.
14448 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
14449 (create_specialized_node): Likewise.
14450 (ipcp_driver): Initialize edge_clone_summaries and do not
14453 2018-06-08 Martin Liska <mliska@suse.cz>
14455 * symbol-summary.h (get): New function.
14456 (call_summary::m_initialize_when_cloning): New class member.
14458 2018-06-08 Martin Liska <mliska@suse.cz>
14460 * cgraph.c (cgraph_node::remove): Do not recycle uid.
14461 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
14462 (symbol_table::allocate_cgraph_symbol): Do not set uid.
14463 * passes.c (uid_hash_t): Record removed_nodes by their uids.
14464 (remove_cgraph_node_from_order): Use the removed_nodes set.
14465 (do_per_function_toporder): Likwise.
14466 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
14467 instead of summary_uid.
14468 (symtab_removal): Likewise.
14469 (symtab_duplication): Likewise.
14471 2018-06-08 Martin Liska <mliska@suse.cz>
14473 * ipa-cp.c (ipcp_store_bits_results): Use
14474 ipcp_transformation_sum.
14475 (ipcp_store_vr_results): Likewise.
14476 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
14478 (ipcp_transformation_initialize): ... this.
14479 (ipa_set_node_agg_value_chain):
14480 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
14481 (write_ipcp_transformation_info): Likewise.
14482 (read_ipcp_transformation_info): Likewise.
14483 (ipcp_update_bits): Likewise.
14484 (ipcp_update_vr): Likewise.
14485 (ipcp_transform_function): Likewise.
14486 * ipa-prop.h: Rename ipcp_transformation_summary to
14487 ipcp_transformation.
14488 (class ipcp_transformation_t): New function summary.
14489 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
14490 (ipa_get_agg_replacements_for_node): Likewise.
14492 2018-06-08 Martin Liska <mliska@suse.cz>
14494 * ipa-pure-const.c (struct funct_state_d): Do it class instead
14496 (class funct_state_summary_t): New function_summary class.
14497 (has_function_state): Remove.
14498 (get_function_state): Likewise.
14499 (set_function_state): Likewise.
14500 (add_new_function): Likewise.
14501 (funct_state_summary_t::insert): New function.
14502 (duplicate_node_data): Remove.
14503 (remove_node_data): Remove.
14504 (funct_state_summary_t::duplicate): New function.
14505 (register_hooks): Create new funct_state_summaries.
14506 (pure_const_generate_summary): Use it.
14507 (pure_const_write_summary): Likewise.
14508 (pure_const_read_summary): Likewise.
14509 (propagate_pure_const): Likewise.
14510 (propagate_nothrow): Likewise.
14511 (dump_malloc_lattice): Likewise.
14512 (propagate_malloc): Likewise.
14513 (execute): Do not register hooks, just remove summary
14515 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
14518 2018-06-08 Martin Liska <mliska@suse.cz>
14520 * ipa-reference.c (remove_node_data): Remove.
14521 (duplicate_node_data): Likewise.
14522 (class ipa_ref_var_info_summary_t): New class.
14523 (class ipa_ref_opt_summary_t): Likewise.
14524 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
14525 (get_reference_optimization_summary): Use
14526 ipa_ref_opt_sum_summaries.
14527 (set_reference_vars_info): Remove.
14528 (set_reference_optimization_summary): Likewise.
14529 (ipa_init): Create summaries.
14530 (init_function_info): Use function summary.
14531 (ipa_ref_opt_summary_t::duplicate): New function.
14532 (ipa_ref_opt_summary_t::remove): New function.
14533 (get_read_write_all_from_node): Fix GNU coding style.
14534 (propagate): Use function summary.
14535 (write_node_summary_p): Fix GNU coding style.
14536 (stream_out_bitmap): Likewise.
14537 (ipa_reference_read_optimization_summary): Use function summary.
14538 (ipa_reference_c_finalize): Do not release hooks.
14540 2018-06-08 Martin Liska <mliska@suse.cz>
14542 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
14543 (analyze_function_body): Extract multiple calls of get_create.
14544 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
14545 * ipa-inline.c (recursive_inlining): Use ::get method.
14546 * ipa-inline.h (estimate_edge_growth): Likewise.
14548 2018-06-08 Martin Liska <mliska@suse.cz>
14550 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
14552 (hsa_function_summary::hsa_function_summary): Use the new enum
14554 (hsa_gpu_implementation_p): Use hsa_summaries::get.
14555 * hsa-gen.c (hsa_get_host_function): Likewise.
14556 (get_brig_function_name): Likewise.
14557 * ipa-hsa.c (process_hsa_functions): Likewise.
14558 (ipa_hsa_write_summary): Likewise.
14559 * symbol-summary.h (symtab_duplication): Use ::get function/
14560 (get): New function.
14562 2018-06-08 Martin Liska <mliska@suse.cz>
14564 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
14566 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
14567 (hsa_register_kernel): Likewise.
14568 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
14569 * hsa-gen.c (hsa_get_host_function): Likewise.
14570 (get_brig_function_name): Likewise.
14571 (generate_hsa): Likewise.
14572 (pass_gen_hsail::execute): Likewise.
14573 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
14574 (devirtualization_time_bonus): Likewise.
14575 (ipcp_propagate_stage): Likewise.
14576 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
14577 (edge_set_predicate): Likewise.
14578 (evaluate_conditions_for_known_args): Likewise.
14579 (evaluate_properties_for_edge): Likewise.
14580 (ipa_fn_summary::reset): Likewise.
14581 (ipa_fn_summary_t::duplicate): Likewise.
14582 (dump_ipa_call_summary): Likewise.
14583 (ipa_dump_fn_summary): Likewise.
14584 (analyze_function_body): Likewise.
14585 (compute_fn_summary): Likewise.
14586 (estimate_edge_devirt_benefit): Likewise.
14587 (estimate_edge_size_and_time): Likewise.
14588 (estimate_calls_size_and_time): Likewise.
14589 (estimate_node_size_and_time): Likewise.
14590 (inline_update_callee_summaries): Likewise.
14591 (remap_edge_change_prob): Likewise.
14592 (remap_edge_summaries): Likewise.
14593 (ipa_merge_fn_summary_after_inlining): Likewise.
14594 (ipa_update_overall_fn_summary): Likewise.
14595 (read_ipa_call_summary): Likewise.
14596 (inline_read_section): Likewise.
14597 (write_ipa_call_summary): Likewise.
14598 (ipa_fn_summary_write): Likewise.
14599 (ipa_free_fn_summary): Likewise.
14600 * ipa-hsa.c (process_hsa_functions): Likewise.
14601 (ipa_hsa_write_summary): Likewise.
14602 (ipa_hsa_read_section): Likewise.
14603 * ipa-icf.c (sem_function::merge): Likewise.
14604 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
14605 (do_estimate_edge_time): Likewise.
14606 (estimate_size_after_inlining): Likewise.
14607 (estimate_growth): Likewise.
14608 (growth_likely_positive): Likewise.
14609 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
14610 (inline_call): Likewise.
14611 * ipa-inline.c (caller_growth_limits): Likewise.
14612 (can_inline_edge_p): Likewise.
14613 (can_inline_edge_by_limits_p): Likewise.
14614 (compute_uninlined_call_time): Likewise.
14615 (compute_inlined_call_time): Likewise.
14616 (want_inline_small_function_p): Likewise.
14617 (edge_badness): Likewise.
14618 (update_caller_keys): Likewise.
14619 (update_callee_keys): Likewise.
14620 (recursive_inlining): Likewise.
14621 (inline_small_functions): Likewise.
14622 (inline_to_all_callers_1): Likewise.
14623 (dump_overall_stats): Likewise.
14624 (early_inline_small_functions): Likewise.
14625 (early_inliner): Likewise.
14626 * ipa-inline.h (estimate_edge_growth): Likewise.
14627 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
14628 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
14629 * ipa-prop.h (IPA_NODE_REF): Likewise.
14630 (IPA_EDGE_REF): Likewise.
14631 * ipa-pure-const.c (malloc_candidate_p): Likewise.
14632 (propagate_malloc): Likewise.
14633 * ipa-split.c (execute_split_functions): Likewise.
14634 * symbol-summary.h: Rename get to get_create.
14636 (get_create): Likewise.
14637 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
14639 2018-06-08 Martin Liska <mliska@suse.cz>
14641 * symbol-summary.h (release): Move definition out of class
14643 (symtab_removal): Likewise.
14644 (symtab_duplication): Likewise.
14646 2018-06-08 Martin Liska <mliska@suse.cz>
14648 * symbol-summary.h (function_summary): Move constructor
14649 implementation out of class declaration.
14650 (release): Likewise.
14651 (symtab_insertion): Likewise.
14652 (symtab_removal): Likewise.
14653 (symtab_duplication): Likewise.
14656 2018-06-08 Martin Liska <mliska@suse.cz>
14658 * Makefile.in: Remove support for MPX (macros, related functions,
14659 fields in cgraph_node, ...).
14660 * builtin-types.def (BT_BND): Likewise.
14661 (BT_FN_BND_CONST_PTR): Likewise.
14662 (BT_FN_CONST_PTR_BND): Likewise.
14663 (BT_FN_VOID_PTR_BND): Likewise.
14664 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
14665 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
14666 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
14667 (expand_builtin_mempcpy_with_bounds): Likewise.
14668 (expand_builtin_memset_with_bounds): Likewise.
14669 (expand_builtin_memset_args): Likewise.
14670 (std_expand_builtin_va_start): Likewise.
14671 (expand_builtin): Likewise.
14672 (expand_builtin_with_bounds): Likewise.
14673 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
14674 (DEF_LIB_BUILTIN_CHKP): Likewise.
14675 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
14676 (DEF_CHKP_BUILTIN): Likewise.
14677 (BUILT_IN_MEMCPY): Likewise.
14678 (BUILT_IN_MEMMOVE): Likewise.
14679 (BUILT_IN_MEMPCPY): Likewise.
14680 (BUILT_IN_MEMSET): Likewise.
14681 (BUILT_IN_STPCPY): Likewise.
14682 (BUILT_IN_STRCAT): Likewise.
14683 (BUILT_IN_STRCHR): Likewise.
14684 (BUILT_IN_STRCPY): Likewise.
14685 (BUILT_IN_STRLEN): Likewise.
14686 (BUILT_IN_MEMCPY_CHK): Likewise.
14687 (BUILT_IN_MEMMOVE_CHK): Likewise.
14688 (BUILT_IN_MEMPCPY_CHK): Likewise.
14689 (BUILT_IN_MEMSET_CHK): Likewise.
14690 (BUILT_IN_STPCPY_CHK): Likewise.
14691 (BUILT_IN_STRCAT_CHK): Likewise.
14692 (BUILT_IN_STRCPY_CHK): Likewise.
14693 * calls.c (store_bounds): Likewise.
14694 (emit_call_1): Likewise.
14695 (special_function_p): Likewise.
14696 (maybe_warn_nonstring_arg): Likewise.
14697 (initialize_argument_information): Likewise.
14698 (finalize_must_preallocate): Likewise.
14699 (compute_argument_addresses): Likewise.
14700 (expand_call): Likewise.
14701 * cfgexpand.c (expand_call_stmt): Likewise.
14702 (expand_return): Likewise.
14703 (expand_gimple_stmt_1): Likewise.
14704 (pass_expand::execute): Likewise.
14705 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
14706 (cgraph_node::remove): Likewise.
14707 (cgraph_node::dump): Likewise.
14708 (cgraph_node::verify_node): Likewise.
14709 * cgraph.h (chkp_function_instrumented_p): Likewise.
14710 (symtab_node::get_alias_target): Likewise.
14711 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
14712 (cgraph_local_p): Likewise.
14713 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
14714 (cgraph_edge::rebuild_references): Likewise.
14715 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
14716 (walk_polymorphic_call_targets): Likewise.
14717 (cgraph_node::expand_thunk): Likewise.
14718 (symbol_table::output_weakrefs): Likewise.
14719 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
14720 (ix86_handle_option): Likewise.
14721 * config/i386/constraints.md: Likewise.
14722 * config/i386/i386-builtin-types.def (BND): Likewise.
14726 * config/i386/i386-builtin.def (BDESC_END): Likewise.
14727 (BDESC_FIRST): Likewise.
14729 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
14730 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
14731 * config/i386/i386.c (enum reg_class): Likewise.
14732 (ix86_target_string): Likewise.
14733 (ix86_option_override_internal): Likewise.
14734 (ix86_conditional_register_usage): Likewise.
14735 (ix86_valid_target_attribute_inner_p): Likewise.
14736 (ix86_set_indirect_branch_type): Likewise.
14737 (ix86_set_current_function): Likewise.
14738 (ix86_function_arg_regno_p): Likewise.
14739 (init_cumulative_args): Likewise.
14740 (ix86_function_arg_advance): Likewise.
14741 (ix86_function_arg): Likewise.
14742 (ix86_pass_by_reference): Likewise.
14743 (ix86_function_value_regno_p): Likewise.
14744 (ix86_function_value_1): Likewise.
14745 (ix86_function_value_bounds): Likewise.
14746 (ix86_return_in_memory): Likewise.
14747 (ix86_setup_incoming_vararg_bounds): Likewise.
14748 (ix86_va_start): Likewise.
14749 (indirect_thunk_need_prefix): Likewise.
14750 (print_reg): Likewise.
14751 (ix86_print_operand): Likewise.
14752 (ix86_expand_call): Likewise.
14753 (ix86_output_function_return): Likewise.
14754 (reg_encoded_number): Likewise.
14755 (BDESC_VERIFYS): Likewise.
14756 (ix86_init_mpx_builtins): Likewise.
14757 (ix86_init_builtins): Likewise.
14758 (ix86_emit_cmove): Likewise.
14759 (ix86_emit_move_max): Likewise.
14760 (ix86_expand_builtin): Likewise.
14761 (ix86_builtin_mpx_function): Likewise.
14762 (ix86_get_arg_address_for_bt): Likewise.
14763 (ix86_load_bounds): Likewise.
14764 (ix86_store_bounds): Likewise.
14765 (ix86_load_returned_bounds): Likewise.
14766 (ix86_store_returned_bounds): Likewise.
14767 (ix86_class_likely_spilled_p): Likewise.
14768 (ix86_hard_regno_mode_ok): Likewise.
14769 (x86_order_regs_for_local_alloc): Likewise.
14770 (ix86_mitigate_rop): Likewise.
14771 (ix86_bnd_prefixed_insn_p): Likewise.
14772 (ix86_mpx_bound_mode): Likewise.
14773 (ix86_make_bounds_constant): Likewise.
14774 (ix86_initialize_bounds): Likewise.
14775 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
14776 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
14777 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
14778 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
14779 (TARGET_CHKP_BOUND_MODE): Likewise.
14780 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
14781 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
14782 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
14783 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
14784 * config/i386/i386.h (TARGET_MPX): Likewise.
14785 (TARGET_MPX_P): Likewise.
14786 (VALID_BND_REG_MODE): Likewise.
14787 (FIRST_BND_REG): Likewise.
14788 (LAST_BND_REG): Likewise.
14789 (enum reg_class): Likewise.
14790 (BND_REG_P): Likewise.
14791 (BND_REGNO_P): Likewise.
14792 (BNDmode): Likewise.
14793 (ADJUST_INSN_LENGTH): Likewise.
14794 * config/i386/i386.md: Likewise.
14795 * config/i386/i386.opt: Likewise.
14796 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
14797 (defined): Likewise.
14798 (LINK_MPX): Likewise.
14799 (MPX_SPEC): Likewise.
14800 (LIBMPX_SPEC): Likewise.
14801 (LIBMPXWRAPPERS_SPEC): Likewise.
14802 (CHKP_SPEC): Likewise.
14803 * config/i386/predicates.md: Likewise.
14804 * dbxout.c (dbxout_type): Likewise.
14805 * doc/extend.texi: Likewise.
14806 * doc/invoke.texi: Likewise.
14807 * doc/md.texi: Likewise.
14808 * doc/tm.texi: Likewise.
14809 * doc/tm.texi.in: Likewise.
14810 * dwarf2out.c (is_base_type): Likewise.
14811 (gen_formal_types_die): Likewise.
14812 (gen_subprogram_die): Likewise.
14813 (gen_type_die_with_usage): Likewise.
14814 (gen_decl_die): Likewise.
14815 (dwarf2out_late_global_decl): Likewise.
14816 * expr.c (expand_assignment): Likewise.
14817 (emit_storent_insn): Likewise.
14818 (store_expr_with_bounds): Likewise.
14819 (store_expr): Likewise.
14820 (expand_expr_real_1): Likewise.
14821 * expr.h (store_expr_with_bounds): Likewise.
14822 * function.c (use_register_for_decl): Likewise.
14823 (struct bounds_parm_data): Likewise.
14824 (assign_parms_augmented_arg_list): Likewise.
14825 (assign_parm_find_entry_rtl): Likewise.
14826 (assign_parm_is_stack_parm): Likewise.
14827 (assign_parm_load_bounds): Likewise.
14828 (assign_bounds): Likewise.
14829 (assign_parms): Likewise.
14830 (expand_function_start): Likewise.
14831 * gcc.c (CHKP_SPEC): Likewise.
14832 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
14833 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
14834 (wrestrict_dom_walker::check_call): Likewise.
14835 * gimple.c (gimple_build_call_from_tree): Likewise.
14836 * gimple.h (enum gf_mask): Likewise.
14837 (gimple_call_with_bounds_p): Likewise.
14838 (gimple_call_set_with_bounds): Likewise.
14839 * gimplify.c (gimplify_init_constructor): Likewise.
14840 * ipa-cp.c (initialize_node_lattices): Likewise.
14841 (propagate_constants_across_call): Likewise.
14842 (find_more_scalar_values_for_callers_subset): Likewise.
14843 * ipa-hsa.c (process_hsa_functions): Likewise.
14844 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
14845 * ipa-icf.c (sem_function::merge): Likewise.
14846 * ipa-inline.c (early_inliner): Likewise.
14847 * ipa-pure-const.c (warn_function_noreturn): Likewise.
14848 (warn_function_cold): Likewise.
14849 (propagate_pure_const): Likewise.
14850 * ipa-ref.h (enum GTY): Likewise.
14851 * ipa-split.c (find_retbnd): Likewise.
14852 (consider_split): Likewise.
14853 (split_function): Likewise.
14854 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
14855 * ipa.c (walk_polymorphic_call_targets): Likewise.
14856 (symbol_table::remove_unreachable_nodes): Likewise.
14857 (process_references): Likewise.
14858 (cgraph_build_static_cdtor_1): Likewise.
14859 * lto-cgraph.c (lto_output_node): Likewise.
14860 (output_refs): Likewise.
14861 (compute_ltrans_boundary): Likewise.
14862 (input_overwrite_node): Likewise.
14863 (input_node): Likewise.
14864 (input_cgraph_1): Likewise.
14865 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
14866 * passes.c (pass_manager::execute_early_local_passes): Likewise.
14867 (class pass_chkp_instrumentation_passes): Likewise.
14868 (make_pass_chkp_instrumentation_passes): Likewise.
14869 * passes.def: Likewise.
14870 * rtl.h (struct GTY): Likewise.
14871 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
14872 * stor-layout.c (layout_type): Likewise.
14873 * symtab.c: Likewise.
14874 * target.def: Likewise.
14875 * targhooks.c (default_chkp_bound_type): Likewise.
14876 (default_chkp_bound_mode): Likewise.
14877 (default_builtin_chkp_function): Likewise.
14878 (default_chkp_function_value_bounds): Likewise.
14879 (default_chkp_make_bounds_constant): Likewise.
14880 (default_chkp_initialize_bounds): Likewise.
14881 * targhooks.h (default_chkp_bound_type): Likewise.
14882 (default_chkp_bound_mode): Likewise.
14883 (default_builtin_chkp_function): Likewise.
14884 (default_chkp_function_value_bounds): Likewise.
14885 (default_chkp_make_bounds_constant): Likewise.
14886 (default_chkp_initialize_bounds): Likewise.
14887 * toplev.c (compile_file): Likewise.
14888 (process_options): Likewise.
14889 * tree-core.h (DEF_BUILTIN): Likewise.
14890 (DEF_BUILTIN_CHKP): Likewise.
14891 * tree-inline.c (declare_return_variable): Likewise.
14892 (remap_gimple_stmt): Likewise.
14893 (copy_bb): Likewise.
14894 (initialize_inlined_parameters): Likewise.
14895 (expand_call_inline): Likewise.
14896 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
14897 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
14898 (make_pass_ipa_chkp_produce_thunks): Likewise.
14899 (make_pass_chkp): Likewise.
14900 (make_pass_chkp_opt): Likewise.
14901 (make_pass_chkp_instrumentation_passes): Likewise.
14902 * tree-pretty-print.c (dump_generic_node): Likewise.
14903 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
14904 * tree-ssa-dce.c (propagate_necessity): Likewise.
14905 (eliminate_unnecessary_stmts): Likewise.
14906 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
14907 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
14908 * tree-ssa-sccvn.h: Likewise.
14909 * tree-ssa-strlen.c (get_string_length): Likewise.
14910 (valid_builtin_call): Likewise.
14911 (adjust_last_stmt): Likewise.
14912 (handle_builtin_strchr): Likewise.
14913 (handle_builtin_strcpy): Likewise.
14914 (handle_builtin_stxncpy): Likewise.
14915 (handle_builtin_memcpy): Likewise.
14916 (handle_builtin_strcat): Likewise.
14917 (strlen_check_and_optimize_stmt): Likewise.
14918 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
14919 * tree-streamer-in.c: Likewise.
14920 * tree-streamer.c (record_common_node): Likewise.
14921 * tree.c (tree_code_size): Likewise.
14922 (wide_int_to_tree_1): Likewise.
14923 (type_contains_placeholder_1): Likewise.
14924 (build_common_tree_nodes): Likewise.
14925 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
14926 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
14927 (POINTER_BOUNDS_P): Likewise.
14928 (BOUNDED_TYPE_P): Likewise.
14929 (BOUNDED_P): Likewise.
14930 (CALL_WITH_BOUNDS_P): Likewise.
14931 (pointer_bounds_type_node): Likewise.
14932 * value-prof.c (gimple_ic): Likewise.
14933 * var-tracking.c (vt_add_function_parameters): Likewise.
14934 * varasm.c (make_decl_rtl): Likewise.
14935 (assemble_start_function): Likewise.
14936 (output_constant): Likewise.
14937 (maybe_assemble_visibility): Likewise.
14938 * varpool.c (ctor_for_folding): Likewise.
14939 * chkp-builtins.def: Remove.
14940 * ipa-chkp.c: Remove.
14941 * ipa-chkp.h: Remove.
14942 * rtl-chkp.c: Remove.
14943 * rtl-chkp.h: Remove.
14944 * tree-chkp-opt.c: Remove.
14945 * tree-chkp.c: Remove.
14946 * tree-chkp.h: Remove.
14948 2018-06-07 Carl Love <cel@us.ibm.com>
14950 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14951 vextract_fp_from_shortl): Add BE support.
14953 2018-06-07 Paul Koning <ni1d@arrl.net>
14955 * compare-elim.c (try_merge_compare): Don't merge compare if
14956 address contains a side effect.
14957 (try_eliminate_compare): Likewise.
14959 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
14961 * config.gcc: Support "tremont".
14962 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
14963 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14965 * config/i386/i386.c (m_TREMONT): Define.
14966 (processor_target_table): Add "tremont".
14967 (PTA_TREMONT): Define.
14968 (ix86_lea_outperforms): Add TARGET_TREMONT.
14969 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
14970 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
14971 and M_INTEL_GOLDMONT_PLUS.
14972 (fold_builtin_cpu): Add "tremont".
14973 (ix86_add_stmt_cost): Add TARGET_TREMONT.
14974 (ix86_option_override_internal): Add "tremont".
14975 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
14976 (processor_type): Add PROCESSOR_TREMONT.
14977 * config/i386/x86-tune.def: Add m_TREMONT.
14978 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
14980 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
14982 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
14983 symbol defined for msp430i* devices to be lower case.
14985 2018-06-07 Richard Biener <rguenther@suse.de>
14987 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
14988 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
14989 Properly wrap signed arithmetic if overflow wraps.
14991 2018-06-07 Jakub Jelinek <jakub@redhat.com>
14993 PR tree-optimization/69615
14994 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
14995 of a cast from a same precision integral SSA_NAME in a bb dominated
14996 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
14997 cast to utype if rhs2 has already a compatible type.
14999 2018-06-07 Richard Biener <rguenther@suse.de>
15001 PR tree-optimization/85935
15002 * graphite-scop-detection.c (find_params_in_bb): Analyze
15003 condition operands with respect to the correct loop. Assert
15004 the analysis doesn't fail.
15006 2018-06-04 Carl Love <cel@us.ibm.com>
15008 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
15009 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
15010 as it is slightly cheaper.
15011 (first_match_or_eos_index_<mode>):
15012 Calculate index using natural element order.
15013 (first_match_index_<mode>):
15014 Calculate index using natural element order.
15015 (first_match_or_eos_index_<mode>):
15016 Calculate index using natural order.
15017 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
15018 for BE and LE modes.
15019 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
15020 P9V_BUILTIN_VCLZLSBB_V16QI.
15021 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
15024 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
15026 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
15027 indentation and line wrap for many prototypes. Add missing
15028 @smallexample directives around block of prototypes for vec_xl and
15031 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
15033 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
15034 track if we pass or return IEEE 128-bit floating point.
15035 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
15036 C++ mangling that is compatible with GCC 8.1.
15037 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
15038 (init_cumulative_args): Note if we pass or return IEEE 128-bit
15039 floating point types.
15040 (rs6000_function_arg_advance_1): Likewise.
15041 (rs6000_mangle_type): Optionally generate mangled names that match
15042 what GCC 8.1 generated for IEEE 128-bit floating point types.
15043 (rs6000_globalize_decl_name): If we have an external function that
15044 passes or returns IEEE 128-bit types, generate a weak reference
15045 from the mangled name used in GCC 8.1 to the current mangled
15047 (rs6000_init_builtins): Make __ibm128 use the long double type if
15048 long double is IBM extended double. Make __float128 use the long
15049 double type if long double is IEEE 128-bit.
15052 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
15053 macro for __ibm128 built-in functions.
15054 (PACK_IF): Add __ibm128 pack/unpack functions.
15055 (UNPACK_IF): Likewise.
15056 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
15057 enable long double built-in functions if long double is IEEE
15058 128-bit floating point.
15059 (rs6000_invalid_builtin): Update long double built-in function
15061 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
15062 functions, adjust the built-in function to use the long double
15063 built-in function if __ibm128 and long double are the same type.
15064 * doc/extend.texi (PowerPC builtins): Update documention for
15065 __builtin_{,un}pack_longdouble. Add documentation for
15066 __builtin_{,un}pack_ibm128.
15068 2018-06-06 Jim Wilson <jimw@sifive.com>
15070 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
15071 (struct machine_function): New field interrupt_mode.
15072 (riscv_handle_type_attribute): New function. Add forward declaration.
15073 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
15074 (riscv_expand_epilogue): Check interrupt_mode field.
15075 (riscv_set_current_function): Check interrupt attribute args and
15076 set interrupt_mode field.
15077 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
15078 (riscv_sret, riscv_uret): New.
15079 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
15080 new arguments to interrupt attribute.
15082 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
15085 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
15086 Don't handle -mcpu=power8 if -mpower9-vector is also used.
15088 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
15090 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
15091 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
15092 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
15093 several redundant entries.
15095 2018-06-06 David Malcolm <dmalcolm@redhat.com>
15097 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
15098 type from "rtx" to "rtx_insn *".
15099 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
15100 for local "call_insn", removing cast.
15101 (ix86_expand_call): Likewise, introducing a "call_insn" local.
15103 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
15105 PR tree-optimization/86066
15106 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
15107 for BIT_INSERT_EXPR stores.
15109 2018-06-06 Richard Biener <rguenther@suse.de>
15111 PR tree-optimization/86062
15112 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
15113 component refs ontop
15114 of to be offsetted base.
15116 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
15118 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
15119 to be static and remove check on interrupt attribute name.
15121 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
15123 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
15124 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
15126 2018-06-05 Steve Ellcey <sellcey@cavium.com>
15129 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
15131 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
15132 Remove second argument, change how error is called.
15133 (aarch64_layout_arg): Remove second argument from
15134 aarch64_err_no_fpadvsimd call.
15135 (aarch64_init_cumulative_args): Ditto.
15136 (aarch64_gimplify_va_arg_expr): Ditto.
15137 * config/aarch64/aarch64.md (mov<mode>): Ditto.
15139 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
15141 * config/i386/i386.md (simple_return_indirect_internal): New expander.
15142 (*simple_return_indirect_internal<mode>): Rename from
15143 simple_return_indirect_internal. Use W mode iterator.
15144 (rstorssp): New expander.
15145 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
15146 (clrssbsy): New expander.
15147 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
15149 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
15151 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
15153 (cmse_check_pointed_object): Likewise.
15155 2018-06-05 Martin Liska <mliska@suse.cz>
15157 PR gcov-profile/47618
15158 * doc/invoke.texi: Document how -fprofile-dir format
15161 2018-06-05 Richard Biener <rguenther@suse.de>
15163 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
15164 removal pretend DOM info isn't available so we do not update
15165 it and only remove edges, not dominated blocks. Actually free
15166 DOM info in case we removed something. Remove unreachable blocks.
15167 (mfb_keep_latches): Work with either DOM info or marked backedges.
15168 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
15169 first. Mark backedges if DOM info isn't available.
15170 (Re-)compute DOM info after cleanup_control_flow_pre.
15172 2018-06-05 Richard Biener <rguenther@suse.de>
15174 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
15175 (locus_discrim_hasher::hash): Adjust.
15176 (locus_discrim_hasher::equal): Likewise.
15177 (next_discriminator_for_locus): Work on line directly.
15178 (same_line_p): Pass in expanded locus1 as well.
15179 (assign_discriminators): Avoid redundant location expansions.
15181 2018-06-05 Richard Biener <rguenther@suse.de>
15183 PR tree-optimization/86046
15184 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
15185 if required after clearing TREE_ADDRESSABLE.
15187 2018-06-05 Richard Biener <rguenther@suse.de>
15189 PR tree-optimization/86047
15190 * tree-ssa-loop.c (for_each_index): Glob handling of all
15191 decls and constants and really handle all of them.
15193 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15196 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
15197 qualifier_void_pointer and qualifier_const_void_pointer.
15198 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
15199 (arm_init_builtins): Handle the above.
15200 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
15201 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
15202 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
15205 2018-06-05 Martin Liska <mliska@suse.cz>
15207 * auto-profile.c (read_autofdo_file): Do not use
15208 gcov_ctr_summary struct.
15209 (afdo_callsite_hot_enough_for_early_inline): Likewise.
15210 * coverage.c (struct counts_entry): Likewise.
15211 (read_counts_file): Read just single summary entry.
15212 (get_coverage_counts): Use gcov_summary struct.
15213 * coverage.h (get_coverage_counts): Likewise.
15214 * gcov-dump.c (dump_working_sets): Likewise.
15215 (tag_summary): Dump just single summary.
15216 * gcov-io.c (gcov_write_summary): Write just histogram
15218 (gcov_read_summary): Read just single summary.
15219 (compute_working_sets): Use gcov_summary struct.
15220 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
15221 of GCOV_COUNTERS_SUMMABLE.
15222 (GCOV_COUNTERS_SUMMABLE): Remove.
15223 (GCOV_FIRST_VALUE_COUNTER): Replace with
15224 GCOV_COUNTER_V_INTERVAL.
15225 (struct gcov_ctr_summary): Remove.
15226 (struct gcov_summary): Directly use fields of former
15228 (compute_working_sets): Use gcov_summary struct.
15229 * gcov.c (read_count_file): Do not use ctrs fields.
15230 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
15232 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
15234 * profile.c: Likewise.
15235 * profile.h: Likewise.
15237 2018-06-05 Martin Liska <mliska@suse.cz>
15239 PR gcov-profile/84846
15240 * gcov.c (output_lines): Print working directory only
15241 in intermediate format.
15243 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
15245 * config/s390/s390-builtin-types.def: Add void function type.
15246 * config/s390/s390-builtins.def: Use the function type for the
15249 2018-06-04 Jim Wilson <jimw@sifive.com>
15251 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
15253 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
15254 and maybe_eh_return. Change regno to unsigned int. Use new args to
15255 handle EH_RETURN_DATA_REGNO registers properly.
15256 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
15257 (riscv_expand_epilogue): Update comment. Change argument name and
15258 type. Update code to use new name and type. Pass new args to
15259 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
15261 * config/riscv/riscv.md (NORMAL_RETURN): New.
15262 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
15263 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
15264 (eh_return): Call gen_eh_return_internal and emit barrier.
15265 (eh_return_internal): Call riscv_expand_epilogue.
15267 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
15269 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
15270 bit_insertion field and declare can_be_merged_into method.
15271 (merged_store_group::can_be_merged_into): New method.
15272 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
15273 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
15274 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
15276 2018-06-04 Richard Biener <rguenther@suse.de>
15278 PR tree-optimization/85955
15279 * builtins.c (fold_builtin_sincos): Convert pointers to
15280 destination to appropriate type before dereferencing.
15282 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
15284 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
15286 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
15288 * expr.c (expand_expr_real_1): Force the operand into memory if
15289 its TYPE_MODE is BLKmode and if there is no integer mode for
15290 the number of bits being extracted.
15292 2018-06-04 Jakub Jelinek <jakub@redhat.com>
15296 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
15297 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
15299 2018-06-04 Richard Biener <rguenther@suse.de>
15301 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
15302 (cleanup_tree_cfg_noloop): ... single caller. Do
15303 start_recording_case_labels later.
15305 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
15307 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
15308 to _IMMINTRIN_H_INCLUDED.
15309 * config/i386/pconfigintrin.h: Ditto.
15310 * config/i386/waitpkgintrin.h: Ditto.
15311 * config/i386/immintrin.h: Add includes for sgxintrin.h,
15312 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
15313 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
15314 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
15315 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
15316 waitpkgintrin.h and cldemoteintrin.h.
15318 2018-06-04 Richard Biener <rguenther@suse.de>
15320 PR tree-optimization/86038
15321 * tracer.c (find_best_successor): Check probability for
15322 being initialized, bail out if not.
15324 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
15327 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
15328 of bits to ignore when comparing architectures.
15330 2018-06-04 Jakub Jelinek <jakub@redhat.com>
15332 PR tree-optimization/69615
15333 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
15334 maximum or minimum of the type, try to merge it also as if
15335 range1 is + [-, x - 1] or + [x + 1, -].
15338 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
15340 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
15342 PR tree-optimization/86034
15343 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
15344 the unsigned bitfield type in a bit insertion sequence if it does not
15345 have a larger precision than the bitfield size.
15346 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
15348 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
15350 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
15352 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
15354 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
15355 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
15356 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
15357 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
15359 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
15361 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
15362 Disable -fdelete-null-pointer-checks for ELF toolchain.
15364 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
15365 Kito Cheng <kito.cheng@gmail.com>
15367 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
15368 (nds32le-*-*, nds32be-*-*): Integrate checking process.
15369 (nds32*-*-*): Add glibc and uclibc conditions.
15370 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
15371 (TARGET_EXCEPT_UNWIND_INFO): Define.
15372 * config/nds32/elf.h: New file.
15373 * config/nds32/linux.h: New file.
15374 * config/nds32/nds32-elf.opt: New file.
15375 * config/nds32/nds32-linux.opt: New file.
15376 * config/nds32/nds32-fp-as-gp.c
15377 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
15378 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
15380 (nds32_asm_file_end): Ditto.
15381 (nds32_print_operand): Ditto.
15382 (nds32_insert_attributes): Ditto.
15383 (nds32_init_libfuncs): New function.
15384 (TARGET_HAVE_TLS): Define.
15385 (TARGET_INIT_LIBFUNCS): Define.
15386 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
15388 (TARGET_ELF): Apply different mcmodel setting.
15389 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
15390 been migrated into elf.h and linux.h files.
15391 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
15392 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
15393 (mcmodel): The content has been migrated into nds32-elf.opt and
15394 nds32-linux.opt files.
15395 * config/nds32/t-elf: New file.
15396 * config/nds32/t-linux: New file.
15398 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
15399 Shiva Chen <shiva0217@gmail.com>
15401 * config/nds32/constants.md (unspec_volatile_element): Add
15402 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
15403 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
15405 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
15406 (make_pass_nds32_fp_as_gp): Declare.
15407 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
15409 (nds32_asm_function_end_prologue): Remove unused asm output.
15410 (nds32_asm_function_begin_epilogue): Remove unused asm output.
15411 (nds32_asm_file_start): Output necessary fp_as_gp information.
15412 (nds32_option_override): Adjust register usage.
15413 (nds32_expand_prologue): Consider fp_as_gp situation.
15414 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
15415 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
15418 (simple_return): Ditto.
15419 (omit_fp_begin): Output special directive for fp_as_gp.
15420 (omit_fp_end): Output special directive for fp_as_gp.
15421 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
15422 mforbid-fp-as-gp): New options.
15424 2018-06-01 Mark Wielaard <mark@klomp.org>
15426 * dwarf2out.c (dwarf2out_finish): Remove generation of
15427 DW_AT_loclists_base.
15429 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
15431 * gimple-ssa-store-merging.c: Include gimple-fold.h.
15432 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
15433 (struct merged_store_group): Add bit_insertion field.
15434 (dump_char_array): Use standard hexadecimal format.
15435 (merged_store_group::merged_store_group): Set bit_insertion to false.
15436 (merged_store_group::apply_stores): Use optimal buffer size. Deal
15437 with BIT_INSERT_EXPR stores. Move up code updating the mask and
15438 also print the mask in the dump file.
15439 (pass_store_merging::gate): Minor tweak.
15440 (imm_store_chain_info::coalesce_immediate): Fix wrong association
15441 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
15442 stores with INTEGER_CST stores.
15443 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
15444 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
15445 and use it throughout. Generate bit insertion sequences if need be.
15446 (pass_store_merging::process_store): Remove redundant condition.
15447 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
15449 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
15451 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
15452 the 128-bit floating point types. Fix function comment.
15454 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15456 * config/aarch64/aarch64-simd.md
15457 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
15459 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
15462 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
15464 PR tree-optimization/85989
15465 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
15467 (backprop::intersect_uses): Check it when deciding whether this
15468 is a backedge reference.
15469 (backprop::process_block): Add each phi to m_visited_phis
15470 after visiting it, then clear it at the end.
15472 2018-06-01 Richard Biener <rguenther@suse.de>
15474 * tree-vectorizer.h (vect_dr_stmt): New function.
15475 (vect_get_load_cost): Adjust.
15476 (vect_get_store_cost): Likewise.
15477 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
15478 Use vect_dr_stmt instead of DR_SMTT.
15479 (vect_record_base_alignments): Likewise.
15480 (vect_calculate_target_alignment): Likewise.
15481 (vect_compute_data_ref_alignment): Likewise and make static.
15482 (vect_update_misalignment_for_peel): Likewise.
15483 (vect_verify_datarefs_alignment): Likewise.
15484 (vector_alignment_reachable_p): Likewise.
15485 (vect_get_data_access_cost): Likewise. Pass down
15486 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
15487 (vect_get_peeling_costs_all_drs): Likewise.
15488 (vect_peeling_hash_get_lowest_cost): Likewise.
15489 (vect_enhance_data_refs_alignment): Likewise.
15490 (vect_find_same_alignment_drs): Likewise.
15491 (vect_analyze_data_refs_alignment): Likewise.
15492 (vect_analyze_group_access_1): Likewise.
15493 (vect_analyze_group_access): Likewise.
15494 (vect_analyze_data_ref_access): Likewise.
15495 (vect_analyze_data_ref_accesses): Likewise.
15496 (vect_vfa_segment_size): Likewise.
15497 (vect_small_gap_p): Likewise.
15498 (vectorizable_with_step_bound_p): Likewise.
15499 (vect_prune_runtime_alias_test_list): Likewise.
15500 (vect_analyze_data_refs): Likewise.
15501 (vect_supportable_dr_alignment): Likewise.
15502 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
15503 (vect_gen_prolog_loop_niters): Likewise.
15504 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
15505 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
15507 (vect_recog_mask_conversion_pattern): Likewise.
15508 (vect_try_gather_scatter_pattern): Likewise.
15509 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
15510 to vect_get_store_cost.
15511 (vect_get_store_cost): Get stmt_info instead of DR.
15512 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
15513 (vect_get_load_cost): Get stmt_info instead of DR.
15515 2018-06-01 Richard Biener <rguenther@suse.de>
15517 PR middle-end/86017
15518 * gimple-fold.c (var_decl_component_p): Also allow offsetted
15519 vars wrapped in MEM_REFs.
15521 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
15523 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
15524 Fix subreg tests so that we only return a choice between
15525 GENERAL_REGS and FP_REGS if the original classes included both.
15527 2018-06-01 Richard Biener <rguenther@suse.de>
15530 * tree-ssa-structalias.c (get_function_part_constraint):
15531 Handle NULL fi->decl.
15532 (find_func_aliases_for_call): Properly handle indirect
15533 fi from direct call.
15534 (find_func_clobbers): Likewise.
15535 (ipa_pta_execute): Likewise.
15536 (create_variable_info_for): For functions that are ifunc_resolver
15537 resolve to a varinfo that contains the result of the resolver call.
15538 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
15541 2018-05-31 Michael Collison <michael.collison@arm.com>
15543 * config/aarch64/aarch64.md:
15544 (*fix_to_zero_extenddfdi2): New pattern.
15545 * gcc.target/aarch64/fix_extend1.c: New testcase.
15547 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
15549 PR middle-end/78809
15550 PR middle-end/83026
15551 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
15552 and BUILT_IN_STRNCMP_EQ.
15553 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
15554 BUILT_IN_STRNCMP_EQ.
15555 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
15556 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
15557 (gimple_fold_builtin): Likewise.
15558 * tree-ssa-strlen.c (compute_string_length): New function.
15559 (determine_min_obsize): New function.
15560 (handle_builtin_string_cmp): New function to handle calls to
15561 string compare functions.
15562 (strlen_optimize_stmt): Add handling to builtin string compare
15564 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
15565 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
15566 * tree.c (build_common_builtin_nodes): Add new defines of
15567 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
15569 2018-05-31 Jakub Jelinek <jakub@redhat.com>
15572 * bb-reorder.c (pass_partition_blocks::gate): Return false for
15573 functions with naked attribute.
15575 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
15577 * config/i386/sse.md (avx_vec_concat<mode>):
15578 Substitute concat_tg_mode mode attribute with xtg_mode.
15579 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
15580 (concat_tg_mode): Remove mode attribute.
15582 2018-05-31 Martin Sebor <msebor@redhat.com>
15585 * calls.c (alloc_max_size): Correct a logic error/typo.
15586 Treat excessive arguments as infinite. Warn for invalid arguments.
15587 * doc/invoke.texi (-Walloc-size-larger-than): Update.
15589 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
15592 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
15593 and movx for Haswell.
15595 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
15596 Cesar Philippidis <cesar@codesourcery.com>
15598 PR middle-end/85879
15599 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
15600 when emitting error on private/firstprivate reductions.
15601 * omp-low.c (lower_omp_target): Avoid reference-type processing
15602 on pointers for firstprivate clause.
15604 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
15606 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
15609 * config/aarch64/aarch64-simd.md
15610 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
15611 (aarch64_ld1_x3_<mode>): Likewise
15612 (aarch64_st1x2<VALLDIF:mode>): Likewise
15613 (aarch64_st1_x2_<mode>): Likewise
15614 (aarch64_st1x3<VALLDIF:mode>): Likewise
15615 (aarch64_st1_x3_<mode>): Likewise
15616 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
15617 (vld1_s8_x3): Likewise.
15618 (vld1_u16_x3): Likewise.
15619 (vld1_s16_x3): Likewise.
15620 (vld1_u32_x3): Likewise.
15621 (vld1_s32_x3): Likewise.
15622 (vld1_u64_x3): Likewise.
15623 (vld1_s64_x3): Likewise.
15624 (vld1_f16_x3): Likewise.
15625 (vld1_f32_x3): Likewise.
15626 (vld1_f64_x3): Likewise.
15627 (vld1_p8_x3): Likewise.
15628 (vld1_p16_x3): Likewise.
15629 (vld1_p64_x3): Likewise.
15630 (vld1q_u8_x3): Likewise.
15631 (vld1q_s8_x3): Likewise.
15632 (vld1q_u16_x3): Likewise.
15633 (vld1q_s16_x3): Likewise.
15634 (vld1q_u32_x3): Likewise.
15635 (vld1q_s32_x3): Likewise.
15636 (vld1q_u64_x3): Likewise.
15637 (vld1q_s64_x3): Likewise.
15638 (vld1q_f16_x3): Likewise.
15639 (vld1q_f32_x3): Likewise.
15640 (vld1q_f64_x3): Likewise.
15641 (vld1q_p8_x3): Likewise.
15642 (vld1q_p16_x3): Likewise.
15643 (vld1q_p64_x3): Likewise.
15644 (vst1_s64_x2): Likewise.
15645 (vst1_u64_x2): Likewise.
15646 (vst1_f64_x2): Likewise.
15647 (vst1_s8_x2): Likewise.
15648 (vst1_p8_x2): Likewise.
15649 (vst1_s16_x2): Likewise.
15650 (vst1_p16_x2): Likewise.
15651 (vst1_s32_x2): Likewise.
15652 (vst1_u8_x2): Likewise.
15653 (vst1_u16_x2): Likewise.
15654 (vst1_u32_x2): Likewise.
15655 (vst1_f16_x2): Likewise.
15656 (vst1_f32_x2): Likewise.
15657 (vst1_p64_x2): Likewise.
15658 (vst1q_s8_x2): Likewise.
15659 (vst1q_p8_x2): Likewise.
15660 (vst1q_s16_x2): Likewise.
15661 (vst1q_p16_x2): Likewise.
15662 (vst1q_s32_x2): Likewise.
15663 (vst1q_s64_x2): Likewise.
15664 (vst1q_u8_x2): Likewise.
15665 (vst1q_u16_x2): Likewise.
15666 (vst1q_u32_x2): Likewise.
15667 (vst1q_u64_x2): Likewise.
15668 (vst1q_f16_x2): Likewise.
15669 (vst1q_f32_x2): Likewise.
15670 (vst1q_f64_x2): Likewise.
15671 (vst1q_p64_x2): Likewise.
15672 (vst1_s64_x3): Likewise.
15673 (vst1_u64_x3): Likewise.
15674 (vst1_f64_x3): Likewise.
15675 (vst1_s8_x3): Likewise.
15676 (vst1_p8_x3): Likewise.
15677 (vst1_s16_x3): Likewise.
15678 (vst1_p16_x3): Likewise.
15679 (vst1_s32_x3): Likewise.
15680 (vst1_u8_x3): Likewise.
15681 (vst1_u16_x3): Likewise.
15682 (vst1_u32_x3): Likewise.
15683 (vst1_f16_x3): Likewise.
15684 (vst1_f32_x3): Likewise.
15685 (vst1_p64_x3): Likewise.
15686 (vst1q_s8_x3): Likewise.
15687 (vst1q_p8_x3): Likewise.
15688 (vst1q_s16_x3): Likewise.
15689 (vst1q_p16_x3): Likewise.
15690 (vst1q_s32_x3): Likewise.
15691 (vst1q_s64_x3): Likewise.
15692 (vst1q_u8_x3): Likewise.
15693 (vst1q_u16_x3): Likewise.
15694 (vst1q_u32_x3): Likewise.
15695 (vst1q_u64_x3): Likewise.
15696 (vst1q_f16_x3): Likewise.
15697 (vst1q_f32_x3): Likewise.
15698 (vst1q_f64_x3): Likewise.
15699 (vst1q_p64_x3): Likewise.
15701 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
15703 * config/msp430/msp430.c (msp430_output_labelref): Prepend
15704 user_label_prefix to name.
15706 * tree-core.h: Update comment about the format of NAME string
15707 passed to handler in attribute_spec.
15709 * config/msp430/msp430.md: Remove erroneous subreg expression from
15710 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
15711 zero_extend{q,h}isi2.
15713 2018-05-30 Borislav Petkov <bp@suse.de>
15715 * doc/extend.texi: Document some architecture specific
15716 constraints and sort entries.
15718 2018-05-30 Martin Sebor <msebor@redhat.com>
15720 PR middle-end/85369
15721 * builtins.c (expand_builtin_stpcpy_1): New function.
15722 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
15723 only if the former succeeds.
15725 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
15727 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
15730 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
15732 * doc/invoke.texi (-flinker-output): Document
15734 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
15736 * passes.c (ipa_write_summaries): Only modify statements if body
15738 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
15739 incrementally linking.
15740 (ipa_passes): Likewise.
15741 * lto-cgraph.c (lto_output_node): When incrementally linking do not
15742 pass down resolution info.
15743 * common.opt (flag_incremental_link): Update info.
15744 * gcc.c (plugin specs): Turn flinker-output=* to
15745 -plugin-opt=-linker-output-known
15746 * toplev.c (compile_file): Also cut compilation when doing incremental
15748 * flag-types. (enum lto_partition_model): Add
15749 LTO_LINKER_OUTPUT_NOLTOREL.
15750 (invoke.texi): Add -flinker-output docs.
15751 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
15752 link same way as WPA; do not stream in dead initializers.
15754 * dwarf2out.c (dwarf2out_die_ref_for_decl,
15755 darf2out_register_external_decl): Support incremental link.
15757 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
15759 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
15761 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
15763 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
15764 it down to simple_object_copy_lto_debug_sections.
15765 (run_gcc): Determine incremental LTO link time and configure
15766 lto1 into non-wpa mode, disable renaming of debug sections.
15768 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
15770 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
15771 descriptions of various incorrectly documented functions.
15773 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
15776 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
15777 address check not strict.
15779 2018-05-30 Richard Biener <rguenther@suse.de>
15781 PR tree-optimization/85964
15782 * tracer.c (better_p): Drop initialized count check, we only
15783 call the function with initialized counts now.
15784 (find_best_successor): Do find a best edge if one
15785 has uninitialized count.
15786 (find_best_predecessor): Likewise. Do BB frequency check only
15787 if count is initialized.
15789 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
15791 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
15792 (aarch64_ldrstr_offset_compare): New.
15793 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
15794 load/store orderings.
15795 (aarch64_gen_adjusted_ldpstp): Likewise.
15797 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
15799 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
15800 Check for subset of GENERAL_REGS and FP_REGS.
15801 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
15804 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
15806 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
15807 and wi::to_poly_offset. Add the current offset and then check
15808 whether the sum fits, rather than using an unchecked addition of
15809 a checked term. Check for a shwi rather than a uhwi.
15810 * expr.c (get_bit_range): Use tree_to_poly_uint64.
15811 (store_constructor): Use poly_int_tree_p.
15812 (expand_expr_real_1): Likewise.
15813 * function.c (assign_temp): Likewise.
15814 * fold-const.c (const_binop): Use poly_int_tree_p and
15815 wi::to_poly_offset.
15816 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
15818 * ipa-icf-gimple.c (func_checker::compare_operand): Use
15819 to_poly_offset for MEM offsets.
15820 * ipa-icf.c (sem_variable::equals): Likewise.
15821 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
15822 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
15823 wi::to_poly_offset for BIT_FIELD_REF offsets.
15824 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
15825 wi::to_poly_offset.
15826 * var-tracking.c (emit_note_insn_var_location): Use
15827 tree_to_poly_uint64.
15829 2018-05-29 Jim Wilson <jimw@sifive.com>
15831 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
15833 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
15836 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
15837 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
15839 (sse4_1_round<mode>2): Use nonimmediate_operand
15840 for operand 1 predicate.
15842 2018-05-29 Martin Sebor <msebor@redhat.com>
15843 Richard Biener <rguenther@suse.de>
15846 * calls.c (get_size_range): Call determine_value_range instead
15847 of get_value_range..
15848 * tree-vrp.h (determine_value_range): Declared new function.
15849 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
15851 2018-05-29 Richard Biener <rguenther@suse.de>
15853 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
15854 sure to use non-pattern stmts for get_earlier_stmt arguments.
15855 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
15856 called on pattern stmts.
15857 (get_later_stmt): Likewise.
15859 2018-05-29 Martin Liska <mliska@suse.cz>
15861 PR gcov-profile/85759
15862 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
15865 2018-05-29 Jakub Jelinek <jakub@redhat.com>
15867 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
15869 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
15873 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
15874 VEC_PACK_FLOAT_EXPR): New tree codes.
15875 * tree-pretty-print.c (op_code_prio): Handle
15876 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
15877 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
15878 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
15879 * tree-inline.c (estimate_operator_cost): Likewise.
15880 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
15881 * fold-const.c (const_binop): Likewise.
15882 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
15883 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
15884 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
15885 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
15886 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
15887 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
15888 * expr.c (expand_expr_real_2): Likewise.
15889 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
15890 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
15891 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
15893 * optabs.c (expand_widen_pattern_expr): For
15894 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
15895 sign from result type rather than operand's type.
15896 (expand_binop_directly): For vec_packu_float_optab and
15897 vec_packs_float_optab allow result type to be different from operand's
15899 * optabs-tree.c (optab_for_tree_code): Handle
15900 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
15901 VEC_PACK_FLOAT_EXPR. Formatting fixes.
15902 * tree-vect-generic.c (expand_vector_operations_1): Handle
15903 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
15904 VEC_PACK_FLOAT_EXPR.
15905 * tree-vect-stmts.c (supportable_widening_operation): Handle
15907 (supportable_narrowing_operation): Handle FLOAT_EXPR.
15908 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
15909 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
15910 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
15911 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
15913 (vec_pack<floatprefix>_float_<mode>): New expander.
15914 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
15916 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
15917 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
15918 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
15919 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
15920 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
15922 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
15923 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
15924 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
15925 VEC_PACK_FLOAT_EXPR): Document.
15927 2018-05-29 Richard Biener <rguenther@suse.de>
15929 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
15931 (stmt_vec_info_vec): Make pointer.
15932 (init_stmt_vec_info_vec): Remove.
15933 (free_stmt_vec_info_vec): Likewise.
15934 (set_stmt_vec_info_vec): New function.
15935 (free_stmt_vec_infos): Likewise.
15936 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
15937 (set_vinfo_for_stmt): Likewise.
15938 (get_earlier_stmt): Likewise.
15939 (get_later_stmt): Likewise.
15940 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
15941 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
15942 (vec_info::~vec_info): Free stmt_vec_infos.
15943 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
15944 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
15945 (pass_slp_vectorize::execute): Likewise.
15946 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
15947 (free_stmt_vec_info_vec): Likewise.
15948 (set_stmt_vec_info_vec): New function.
15949 (free_stmt_vec_infos): Likewise.
15950 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
15951 the global stmt_vec_info_vec.
15952 * tree-parloops.c (gather_scalar_reductions): Use
15953 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
15956 2018-05-29 Richard Biener <rguenther@suse.de>
15958 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
15960 2018-05-29 Martin Liska <mliska@suse.cz>
15961 David Malcolm <dmalcolm@redhat.com>
15963 * vec.c (test_reverse): New.
15964 (vec_c_tests): Add new test.
15965 * vec.h (vl_ptr>::reverse): New function.
15967 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
15969 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
15971 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
15974 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
15976 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
15978 2018-05-28 Richard Biener <rguenther@suse.de>
15980 PR tree-optimization/85933
15981 * tree-vect-data-refs.c (vect_record_base_alignments): Only
15982 look at stmts marked as vectorizable.
15984 2018-05-28 Richard Biener <rguenther@suse.de>
15986 PR tree-optimization/85934
15987 * tree-vect-generic.c (expand_vector_operations_1): Hoist
15988 vector boolean check before scalar optimization.
15990 2018-05-28 Jakub Jelinek <jakub@redhat.com>
15992 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
15995 2018-05-28 Mark Wielaard <mark@klomp.org>
15997 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
15998 if it is an expression containing a minus sign.
16000 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
16002 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
16004 2018-05-27 Paul Koning <ni1d@arrl.net>
16006 * config/pdp11/pdp11.md (truncsihi2): Remove.
16008 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
16009 Chung-Ju Wu <jasonwucj@gmail.com>
16011 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
16013 (unaligned_store_dw): Ditto.
16014 * config/nds32/nds32-memory-manipulation.c
16015 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
16016 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
16017 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
16018 (emit_setmem_word_loop): Rename to ...
16019 (emit_setmem_doubleword_loop): ... this.
16020 (nds32_gen_dup_4_byte_to_word_value): New function.
16021 (nds32_gen_dup_8_byte_to_double_word_value): New function.
16022 (nds32_expand_setmem_loop): Refine implementation.
16023 (nds32_expand_setmem_loop_v3m): Ditto.
16024 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
16027 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
16029 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
16031 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
16033 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
16034 (nds32_init_machine_status): Initialize machine->attr_naked_p and
16035 machine->attr_no_prologue_p.
16036 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
16037 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
16038 (nds32_expand_epilogue): Consider attr_naked_p.
16039 (nds32_expand_epilogue_v3pop): Likewise.
16040 (nds32_can_use_return_insn): Likewise.
16041 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
16042 attr_no_prologue_p fields.
16043 * config/nds32/nds32.opt (mret-in-naked-func): New option.
16045 2018-05-27 Jakub Jelinek <jakub@redhat.com>
16048 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
16050 * config/i386/sse.md
16051 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
16053 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
16055 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
16057 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
16059 (*<floatsuffix>floatv2div2sf2): Rename to ...
16060 (*float<floatunssuffix>v2div2sf2): ... this.
16061 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
16062 (float<floatunssuffix>v2div2sf2_mask): ... this.
16063 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
16064 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
16065 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
16067 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
16069 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
16071 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
16073 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
16075 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
16077 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
16078 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
16079 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
16080 gen_ufix_truncv8dfv8si2.
16081 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
16082 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
16083 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
16084 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
16085 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
16086 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
16087 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
16088 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
16090 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
16094 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
16096 2018-05-25 Jim Wilson <jimw@sifive.com>
16098 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
16099 * config/riscv/riscv.c (struct machine_function): Add
16100 interrupt_handler_p and attribute_checked_p fields.
16101 (riscv_attribute_table): Add interrupt.
16102 (riscv_interrupt_type_p): New.
16103 (riscv_save_reg_p): Save extra regs for interrupt handler.
16104 (riscv_use_save_libcall): Return false for interrupt handler.
16105 (riscv_first_stack_step): Add forward declaration.
16106 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
16107 for interrupt handler with large frame. Use it for saved reg list.
16108 (riscv_expand_prologue): Move flag_stack_usage_info support to
16109 eliminate duplication.
16110 (riscv_expand_epilogue): Generate mret for interrupt handler.
16111 (riscv_epilogue_uses): New.
16112 (riscv_can_use_return_insn): Return false for interrupt handler.
16113 (riscv_function_ok_for_sibcall): Likewise.
16114 (riscv_set_current_function): Add interrupt handler support.
16115 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
16116 * config/riscv/riscv.md (UNSPECV_MRET): New.
16118 (riscv_frflags, riscv_fsflags): Use tab after opcode.
16120 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
16122 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
16124 PR tree-optimization/85712
16125 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
16126 this candidate has already been replaced in-situ by a copy.
16128 2018-05-25 Jason Merrill <jason@redhat.com>
16130 PR c++/80485 - inline function non-zero address.
16131 * symtab.c (nonzero_address): Check DECL_COMDAT.
16133 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
16136 * config/alpha/alpha.md (ashlsi3): New insn pattern.
16137 (*ashlsi_se): Rename from *ashldi_se. Define as sign
16138 extension of SImode operation. Use const123_operand predicate.
16139 (*saddsi_1): Remove.
16140 (*saddl_se_1): Ditto.
16141 (*ssubsi_1): Ditto.
16142 (*ssubl_se_1): Ditto.
16143 * config/alpha/predicates.md (const123_operand): New predicate.
16144 * config/alpha/constraints.md (P): Use IN_RANGE.
16146 2018-05-25 Richard Biener <rguenther@suse.de>
16148 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
16150 (ref_maybe_used_by_stmt_p): Likewise.
16151 (stmt_may_clobber_ref_p): Likewise.
16152 (stmt_may_clobber_ref_p_1): Likewise.
16153 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
16155 (ref_maybe_used_by_stmt_p): Likewise.
16156 (stmt_may_clobber_ref_p): Likewise.
16157 (stmt_may_clobber_ref_p_1): Likewise.
16158 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
16159 the alias oracle to disambiguate DRs with stmts DR analysis
16161 (vect_analyze_data_refs): Do not give up on not analyzable
16162 DRs for BB vectorization. Remove code truncating the dataref
16165 2018-05-25 Jakub Jelinek <jakub@redhat.com>
16168 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
16169 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
16170 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
16172 2018-05-25 Richard Biener <rguenther@suse.de>
16174 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
16175 function, combining stmt data ref gathering and fatal analysis
16177 (vect_analyze_data_refs): Remove now redudnant code and simplify.
16178 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
16179 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
16180 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
16181 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
16183 2018-05-25 Bin Cheng <bin.cheng@arm.com>
16185 PR tree-optimization/85720
16186 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
16187 SCC if all partitions are builtins.
16188 (version_loop_by_alias_check): New parameter. Generate cancelable
16189 runtime alias check if all partitions are builtins.
16190 (distribute_loop): Update call to above function.
16192 2018-05-25 Bin Cheng <bin.cheng@arm.com>
16194 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
16195 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
16196 (parm_default_def_partition_arg): Ditto.
16197 (set_parm_default_def_partition): Ditto.
16198 (get_parm_default_def_partitions): Ditto and make it static.
16199 (get_undefined_value_partitions): Ditto and make it static.
16200 (remove_ssa_form): Refactor call to init_var_map here.
16201 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
16202 computation for loop region.
16203 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
16204 (register_default_def): Delete.
16205 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
16206 (parm_default_def_partition_arg): Ditto.
16207 (set_parm_default_def_partition): Ditto.
16208 (get_parm_default_def_partitions): Ditto and make it static.
16209 (get_undefined_value_partitions): Ditto and make it static.
16210 (coalesce_with_default, coalesce_with_default): Update comment.
16211 (create_coalesce_list_for_region): New func factored out from
16212 create_outofssa_var_map.
16213 (populate_coalesce_list_for_outofssa): New func factored out from
16214 create_outofssa_var_map and coalesce_ssa_name.
16215 (create_outofssa_var_map): Delete.
16216 (coalesce_ssa_name): Refactor to support live range computation.
16217 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
16218 (get_parm_default_def_partitions): Delete.
16219 (get_undefined_value_partitions): Ditto.
16220 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
16221 computation for loop region.
16222 (new_tree_live_info, loe_visit_block): Ditto.
16223 (live_worklist, set_var_live_on_entry): Ditto.
16224 (calculate_live_on_exit, verify_live_on_entry): Ditto.
16225 * tree-ssa-live.h (struct _var_map): New fields.
16226 (init_var_map): Change decl.
16227 (region_contains_p): New.
16229 2018-05-25 Bin Cheng <bin.cheng@arm.com>
16231 * tree-ssa-live.h (live_merge_and_clear): Delete.
16233 2018-05-25 Richard Biener <rguenther@suse.de>
16236 * tree-dump.c (dequeue_and_dump): Remove access to removed
16237 operand 2 of a SWITCH_EXPR.
16239 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
16241 * doc/sourcebuild.texi (vect_double_cond_arith): Include
16242 multiplication and division.
16243 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
16244 (cond_udiv@var{m}, cond_umod@var{m}): Document.
16245 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
16246 (cond_udiv_optab, cond_umod_optab): New optabs.
16247 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
16248 (IFN_COND_RDIV): New internal functions.
16249 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
16250 TRUNC_MOD_EXPR and RDIV_EXPR.
16251 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
16252 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
16254 (SVE_INT_BINARY): Include mult.
16255 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
16256 (optab, sve_int_op): Handle mult.
16257 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
16259 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
16260 for SVE_INT_BINARY_SD.
16262 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
16264 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
16265 (optab, sve_int_op): Handle div and udiv.
16266 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
16267 for SVE_INT_BINARY_SD.
16268 (*<optab><mode>3): New insn for the same.
16270 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
16272 * tree-vect-patterns.c: Include predict.h.
16273 (vect_recog_divmod_pattern): Restrict check for division support
16274 to when optimizing for size.
16276 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
16278 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
16279 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
16280 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
16281 (gimple_match_op::set_op): Likewise.
16282 (gimple_resimplify4): Declare.
16283 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
16284 (expr::gen_transform): Likewise.
16285 (decision_tree::gen): Generate a simplification routine for 4 operands.
16286 * gimple-match-head.c (gimple_simplify): Add an overload for
16287 4 operands. In the top-level function, handle up to 4 call
16288 arguments and call gimple_resimplify4.
16289 (gimple_resimplify4): New function.
16290 (build_call_internal): Pass a fourth operand.
16291 (maybe_push_to_seq): Likewise.
16292 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
16293 Fold VEC_COND_EXPRs of an operation and a default value into
16294 an IFN_COND_* function if possible.
16295 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
16297 (SVE_COND_FP_BINARY): Include them.
16298 (optab, sve_fp_op): Handle them.
16299 (SVE_INT_BINARY_REV): New code iterator.
16300 (SVE_COND_FP_BINARY_REV): New int iterator.
16301 (commutative): New int attribute.
16302 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
16304 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
16306 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
16307 (*cond_<optab><mode>): New patterns for reversed operands.
16309 2018-05-25 Richard Biener <rguenther@suse.de>
16311 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
16312 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
16313 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
16314 (STMT_VINFO_GROUPED_ACCESS): Adjust.
16315 * tree-vect-data-refs.c (everywhere): Adjust users.
16316 * tree-vect-loop.c (everywhere): Likewise.
16317 * tree-vect-slp.c (everywhere): Likewise.
16318 * tree-vect-stmts.c (everywhere): Likewise.
16319 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
16321 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16323 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
16325 (gcc_cv_as_section_exclude): ... this.
16326 Try Solaris as #exclude syntax.
16327 * configure: Regenerate.
16328 * config.in: Regenerate.
16329 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
16331 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
16332 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
16334 * varasm.c (default_elf_asm_named_section): Don't check if
16335 HAVE_GAS_SECTION_EXCLUDE is defined.
16337 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
16339 * doc/md.texi: Update the documentation of the cond_* optabs
16340 to mention the new final operand. Fix GET_MODE_NUNITS call.
16341 Describe the scalar case too.
16342 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
16343 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
16345 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
16346 (get_conditional_internal_fn): Update comment.
16347 * tree-vect-loop.c (vectorizable_reduction): Pass the original
16348 accumulator value as a final argument to conditional functions.
16349 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
16350 a define_expand and add an "else" operand. Assert for now that
16351 the else operand is equal to operand 2. Use SVE_INT_BINARY and
16352 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
16353 (*cond_<optab><mode>): New patterns.
16354 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
16355 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
16356 (UNSPEC_COND_EOR): Delete.
16357 (optab): Remove associated mappings.
16358 (SVE_INT_BINARY): New code iterator.
16359 (sve_int_op): Remove int attribute and add "minus" to the code
16361 (SVE_COND_INT_OP): Delete.
16362 (SVE_COND_FP_OP): Rename to...
16363 (SVE_COND_FP_BINARY): ...this.
16365 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
16367 * optabs.c (can_reuse_operands_p): New function.
16368 (maybe_legitimize_operands): Try to reuse the results for
16371 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
16373 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
16374 Add {q} suffix to insn mnemonic.
16376 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
16378 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
16379 (msp430_warn_func_return): New.
16381 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
16383 * fold-const.c (tree_nonzero_bits): New function.
16384 * fold-const.h (tree_nonzero_bits): Likewise.
16385 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
16386 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
16388 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
16392 * varasm.c (assemble_alias): Check ifunc_resolver only on
16395 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
16398 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
16399 when memory input operand is handled.
16401 2018-05-24 Luis Machado <luis.machado@linaro.org>
16403 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
16405 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
16407 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
16409 * match.pd: Delay FMA folds until after vectorization.
16411 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
16414 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
16415 address check not strict.
16417 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
16419 * gimple-match.h (gimple_match_op): New class.
16420 (mprts_hook): Replace parameters with a gimple_match_op *.
16421 (maybe_build_generic_op): Likewise.
16422 (gimple_simplified_result_is_gimple_val): Replace parameters with
16423 a const gimple_match_op *.
16424 (gimple_simplify): Replace code_helper * and tree * parameters with
16425 a gimple_match_op * parameter.
16426 (gimple_resimplify1): Replace code_helper *, tree and tree *
16427 parameters with a gimple_match_op * parameter.
16428 (gimple_resimplify2): Likewise.
16429 (gimple_resimplify3): Likewise.
16430 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
16431 parameters with a gimple_match_op * parameter.
16432 * gimple-match-head.c (gimple_simplify): Change prototypes of
16433 auto-generated functions to take a gimple_match_op * instead of
16434 separate code_helper * and tree * parameters. Make the same
16435 change in the top-level overload and update calls to the
16436 gimple_resimplify routines. Update calls to the auto-generated
16437 functions and to maybe_push_res_to_seq in the publicly-facing
16438 operation-specific gimple_simplify overloads.
16439 (gimple_match_op::MAX_NUM_OPS): Define.
16440 (gimple_resimplify1): Replace rcode and ops with a single res_op
16441 parameter. Update call to gimple_simplify.
16442 (gimple_resimplify2): Likewise.
16443 (gimple_resimplify3): Likewise.
16444 (mprts_hook): Replace parameters with a gimple_match_op *.
16445 (maybe_build_generic_op): Likewise.
16446 (build_call_internal): Replace type, nargs and ops with
16447 a gimple_match_op *.
16448 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
16449 with a single gimple_match_op *. Update calls to mprts_hook,
16450 build_call_internal and gimple_simplified_result_is_gimple_val.
16451 Factor out code that is common to the tree_code and combined_fn cases.
16452 * genmatch.c (expr::gen_transform): Replace tem_code and
16453 tem_ops with a gimple_match_op called tem_op. Update calls
16454 to the gimple_resimplify functions and maybe_push_res_to_seq.
16455 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
16456 res_ops. Update call to the gimple_resimplify functions.
16457 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
16458 (decision_tree::gen): Make the functions take a gimple_match_op *
16459 called res_op instead of separate res_code and res_ops parameters.
16460 Update call accordingly.
16461 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
16462 and ops with a single res_op parameter. Update calls to
16463 maybe_build_generic_op and maybe_push_res_to_seq.
16464 (fold_stmt_1): Update calls to gimple_simplify and
16465 replace_stmt_with_simplification.
16466 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
16467 and gimple_simplified_result_is_gimple_val.
16468 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
16470 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
16471 with a gimple_match_op *.
16472 (vn_nary_build_or_lookup): Likewise. Update call to
16473 vn_nary_build_or_lookup_1.
16474 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
16475 gimple_match_op *. Update calls to the gimple_resimplify routines
16476 and to gimple_simplified_result_is_gimple_val.
16477 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
16478 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
16479 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
16480 (visit_nary_op): Likewise.
16481 (visit_reference_op_load): Likewise.
16483 2018-05-23 Luis Machado <luis.machado@linaro.org>
16485 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
16486 modifier for printing the step amount.
16488 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16491 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
16494 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
16496 * doc/sourcebuild.texi (Endianness): New subsubsection.
16498 2018-05-23 Luis Machado <luis.machado@linaro.org>
16500 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
16501 <prefetch_dynamic_strides>: New const bool field.
16502 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
16503 prefetch_dynamic_strides.
16504 (exynosm1_prefetch_tune): Likewise.
16505 (thunderxt88_prefetch_tune): Likewise.
16506 (thunderx_prefetch_tune): Likewise.
16507 (thunderx2t99_prefetch_tune): Likewise.
16508 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
16510 (aarch64_override_options_internal): Update to set
16511 PARAM_PREFETCH_DYNAMIC_STRIDES.
16512 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
16513 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
16514 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
16515 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
16516 prefetch-dynamic-strides setting.
16518 2018-05-23 Luis Machado <luis.machado@linaro.org>
16520 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
16521 <minimum_stride>: New const int field.
16522 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
16523 minimum_stride field defaulting to -1.
16524 (exynosm1_prefetch_tune): Likewise.
16525 (thunderxt88_prefetch_tune): Likewise.
16526 (thunderx_prefetch_tune): Likewise.
16527 (thunderx2t99_prefetch_tune): Likewise.
16528 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
16529 <default_opt_level>: Set to 3.
16530 (aarch64_override_options_internal): Update to set
16531 PARAM_PREFETCH_MINIMUM_STRIDE.
16532 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
16533 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
16534 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
16535 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
16536 stride is constant and is below the minimum stride threshold.
16538 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16540 * config/arm/arm-cpus.in (mode26): Delete.
16541 (armv4): Delete mode26 reference.
16542 * config/arm/arm.c (arm_configure_build_target): Delete use of
16545 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
16547 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
16549 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
16550 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
16552 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
16553 (floatunsdidf2): Ditto.
16555 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
16557 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
16558 (fixuns_trunc<mode>si2_avx512f): Ditto.
16559 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
16560 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
16561 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
16563 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
16565 PR rtl-optimization/79985
16566 * df-scan.c (df_insn_refs_collect): Remove special case for
16567 global registers and asm statements.
16569 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
16571 * extend.texi (Global Register Variables): Rewrite the bullet list.
16572 Note that the register is available for allocation. Note that access
16573 via inline asm must use constraints. Add note about async-signal
16574 handlers. Remove paragraph about automagic register selection.
16576 2018-05-23 Richard Biener <rguenther@suse.de>
16578 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
16579 of fixed offset from memset VN.
16581 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
16583 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
16584 first_interp field.
16585 (alloc_cand_and_find_basis): Initialize first_interp field.
16586 (slsr_process_mul): Modify first_interp field.
16587 (slsr_process_add): Likewise.
16588 (slsr_process_cast): Modify first_interp field for each new
16590 (slsr_process_copy): Likewise.
16591 (dump_candidate): Dump first_interp field.
16592 (replace_mult_candidate): Process all interpretations, not just
16594 (replace_rhs_if_not_dup): Likewise.
16595 (replace_one_candidate): Likewise.
16597 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
16599 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
16601 (aarch64_needs_frame_chain): New function.
16602 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
16604 2018-05-23 Sudakshina Das <sudi.das@arm.com>
16607 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
16608 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
16609 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
16610 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
16611 as true for strict-align.
16612 (aarch64_can_inline_p): Perform checks even when callee has no
16613 attributes to check for strict alignment.
16614 * doc/extend.texi (AArch64 Function Attributes): Document
16616 * doc/invoke.texi: (AArch64 Options): Likewise.
16618 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
16620 PR tree-optimization/85853
16621 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
16622 the handling of the root of the node to...
16623 (vect_slp_analyze_node_operations_1): ...this new function,
16624 and run the whole thing with the child nodes' def types
16625 set according to their SLP node's def type.
16627 2018-05-23 Richard Biener <rguenther@suse.de>
16629 PR middle-end/85874
16630 * tree-data-ref.c (create_runtime_alias_checks): Defer
16631 and ignore overflow warnings.
16633 2018-05-23 Yury Gribov <tetra2005@gmail.com>
16635 PR tree-optimization/85822
16636 * tree-vrp.c (is_masked_range_test): Fix handling of negative
16639 2018-05-23 Richard Biener <rguenther@suse.de>
16641 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
16642 memset constants via native_interpret_expr.
16644 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
16647 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
16649 (cgraph_node::create_alias): Likewise.
16650 (cgraph_node::get_availability): Check ifunc_resolver instead
16651 of looking up ifunc attribute.
16652 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
16653 * varasm.c (do_assemble_alias): Likewise.
16654 (assemble_alias): Likewise.
16655 (default_binds_local_p_3): Likewise.
16656 * cgraph.h (cgraph_node): Add ifunc_resolver.
16657 (cgraph_node::only_called_directly_or_aliased_p): Return false
16658 for IFUNC resolver.
16659 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
16661 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
16662 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
16663 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
16664 instead of looking up ifunc attribute.
16666 2018-05-22 Luis Machado <luis.machado@linaro.org>
16668 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
16670 2018-05-22 Martin Sebor <msebor@redhat.com>
16672 PR middle-end/85359
16673 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
16674 only when expasion succeeds.
16675 (expand_builtin_strcmp): Same.
16676 (expand_builtin_strncmp): Same.
16678 2018-05-22 Martin Sebor <msebor@redhat.com>
16680 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
16682 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
16683 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16685 * config/aarch64/aarch64-ldpstp.md: Replace uses of
16686 aarch64_mem_pair_operand with memory_operand and delete operand swapping
16688 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
16689 Add check for legitimate_address.
16690 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
16691 (aarch64_swap_ldrstr_operands): New.
16692 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
16695 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
16696 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16698 * config/aarch64/aarch64.md: New patterns to generate stp
16700 (store_pair_sw, store_pair_dw): New patterns to generate stp for
16701 single words and double words.
16702 (load_pair_sw, load_pair_dw): Likewise.
16703 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
16705 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
16707 * config/aarch64/aarch64-ldpstp.md: Modify peephole
16708 for different mode ldpstp and add peephole for merged zero stores.
16709 Likewise for loads.
16710 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
16712 (aarch64_gen_store_pair): Rename calls to match new patterns.
16713 (aarch64_gen_load_pair): Rename calls to match new patterns.
16714 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
16715 (load_pair<DREG:mode><DREG2:mode>): ... This.
16716 (store_pair<mode>): Rename to...
16717 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
16718 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
16719 New mode iterators.
16720 (V_INT_EQUIV): Handle SImode.
16721 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
16724 2018-05-22 Martin Sebor <msebor@redhat.com>
16727 * calls.c (maybe_warn_nonstring_arg): Use string length to set
16728 or ajust the presumed bound on an operation to avoid unnecessary
16731 2018-05-22 Martin Sebor <msebor@redhat.com>
16733 PR tree-optimization/85826
16734 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
16735 assuming that a DECL necesarily has a constant size.
16737 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
16739 PR middle-end/85862
16740 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
16742 2018-05-22 Richard Biener <rguenther@suse.de>
16744 PR tree-optimization/85834
16745 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
16746 non-constant and non-zero memset arguments.
16748 2018-05-22 Martin Liska <mliska@suse.cz>
16751 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
16753 2018-05-22 Richard Biener <rguenther@suse.de>
16755 PR tree-optimization/85863
16756 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
16757 comparisons when vectype is specified.
16758 (vectorizable_condition): Do not specify vectype for
16759 vect_is_simple_cond when SLP vectorizing.
16761 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
16764 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
16765 define __ibm128 as long double.
16766 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
16767 as a distinct type when IEEE 128-bit support is enabled.
16768 (init_float128_ieee): Fix up conversions between IFmode and IEEE
16769 128-bit types to use the correct functions.
16770 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
16771 convert between 128-bit floating point types that have different
16772 modes but the same representation, instead of using gen_lowpart to
16774 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
16776 (IFKF_reg): New attributes to give the register constraints for
16778 (extend<mode>tf2_internal): New insns to mark an explicit
16779 conversion between 128-bit floating point types that have a
16780 different mode but share the same representation.
16782 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
16784 PR tree-optimization/85814
16785 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
16786 a null return from get_strinfo when unsharing the next
16787 strinfo in the chain.
16789 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
16792 * varasm.c (weak_finish): Clean up weak_decls.
16794 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16796 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
16797 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
16798 UNSPEC_UADALP values.
16799 * config/aarch64/iterators.md (ABAL): New int iterator.
16802 (sur): Add mappings for the above.
16803 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
16805 (aarch64_<sur>abal<mode>_4): Likewise.
16806 (aarch64_<sur>adalp<mode>_3): Likewise.
16807 (<sur>sadv16qi): New define_expand.
16809 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
16811 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
16812 (*movdf_internal): Ditto.
16813 (*rcpsf2_sse): Ditto.
16814 (*rsqrtsf2_sse): Ditto.
16815 (*sqrt<mode>2_sse): Ditto.
16817 2018-05-21 Tamar Christina <tamar.christina@arm.com>
16819 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
16821 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
16822 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
16823 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
16824 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
16826 * config/aarch64/arm_neon.h: Likewise.
16827 * config/aarch64/iterators.md (VQ_I): New.
16829 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
16831 * config.gcc: Add arc/t-multilib-linux to tmake_file for
16833 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
16836 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
16838 * config/nds32/constraints.md (S): New constraint.
16839 * config/nds32/nds32.md (call_internal): Use constraint S.
16840 (call_value_internal): Likewise.
16841 (sibcall_internal): Likewise.
16842 (sibcall_value_internal): Likewise.
16844 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
16845 Chung-Ju Wu <jasonwucj@gmail.com>
16847 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
16848 into consideration.
16850 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
16851 Chung-Ju Wu <jasonwucj@gmail.com>
16853 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
16854 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
16855 (nds32_rtx_costs_impl): Simplify.
16856 (nds32_address_cost_impl): Simplify.
16857 (nds32_init_rtx_costs): New function.
16858 (nds32_rtx_costs_speed_prefer): Likewise.
16859 (nds32_rtx_costs_size_prefer): Likewise.
16860 (nds32_address_cost_speed_prefer): Likewise.
16861 (nds32_address_cost_speed_fwprop): Likewise.
16862 (nds32_address_cost_size_prefer): Likewise.
16863 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
16864 * config/nds32/nds32.c (nds32_option_override): Use
16865 nds32_init_rtx_costs function.
16867 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
16869 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
16870 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
16871 (TARGET_PIPELINE_N8): Likewise.
16872 (TARGET_PIPELINE_N10): Likewise.
16873 (TARGET_PIPELINE_N13): Likewise.
16874 (TARGET_PIPELINE_GRAYWOLF): Likewise.
16876 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
16878 * config/nds32/nds32-fpu.md: Update copyright year.
16880 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16882 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
16884 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16886 * config/nds32/nds32.c
16887 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
16888 * config/nds32/nds32.opt (minline-asm-r15): New option.
16890 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16892 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
16894 * config/nds32/nds32.md (abssi2): New pattern.
16896 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
16898 * config/i386/i386.md (rex64namesuffix): New mode attribute.
16899 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
16900 Merge insn pattern from sse_cvtsi2ss<round_name> and
16901 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
16902 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
16903 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
16904 using SWI48 mode iterator.
16905 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
16906 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
16907 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
16908 pattern from sse_cvttss2si<round_saeonly_name>
16909 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
16910 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
16911 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
16912 using SWI48 mode iterator.
16913 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
16914 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
16915 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
16916 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
16917 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
16918 using SWI48 mode iterator.
16919 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
16920 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
16921 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
16922 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
16923 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
16924 SWI48 mode iterator.
16925 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
16926 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
16927 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
16928 pattern from sse_cvttsd2si<round_saeonly_name>
16929 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
16931 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16933 * config/nds32/nds32-md-auxiliary.c
16934 (nds32_valid_smw_lwm_base_p): Refine.
16935 (nds32_output_smw_single_word): Refine.
16936 (nds32_output_smw_double_word): New.
16937 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
16939 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16941 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
16942 (nds32_output_stack_pop): Refine.
16943 (nds32_expand_unaligned_load): Refine.
16944 (nds32_expand_unaligned_store): Refine.
16946 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
16947 Chung-Ju Wu <jasonwucj@gmail.com>
16949 * config/nds32/constants.md: Add TP_REGNUM constant.
16950 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
16951 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
16953 * config/nds32/nds32-doubleword.md: Consider flag_pic.
16954 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
16955 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
16956 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
16957 and PIC code generation.
16958 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
16960 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
16962 * config/nds32/nds32.md: Support TLS and PIC.
16963 * config/nds32/nds32.c: Support TLS and PIC.
16964 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
16965 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
16968 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16970 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
16971 mode with E_ prefix.
16973 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
16974 Chung-Ju Wu <jasonwucj@gmail.com>
16976 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
16977 * config/nds32/nds32-md-auxiliary.c
16978 (symbolic_reference_mentioned_p): New.
16979 (nds32_legitimize_ict_address): New.
16980 (nds32_expand_ict_move): New.
16981 (nds32_indirect_call_referenced_p): New.
16982 (nds32_symbol_binds_local_p): Delete.
16983 (nds32_long_call_p): Modify.
16984 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
16985 * config/nds32/nds32-protos.h
16986 (symbolic_reference_mentioned_p): Declare.
16987 (nds32_legitimize_ict_address): Declare.
16988 (nds32_expand_ict_move): Declare.
16989 (nds32_indirect_call_referenced_p): Declare.
16990 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
16991 (nds32_relax_group): Use nds32_ict_const_p as condition.
16992 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
16993 (nds32_asm_file_start): Output ict_model directive in asm code.
16994 (nds32_legitimate_address_p): Consider indirect call.
16995 (nds32_print_operand): Consider indirect call.
16996 (nds32_print_operand_address): Consider indirect call.
16997 (nds32_insert_attributes): Handle "indirect_call" attribute.
16998 (TARGET_LEGITIMATE_ADDRESS_P): Define.
16999 (TARGET_LEGITIMATE_CONSTANT_P): Define.
17000 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
17001 (TARGET_DELEGITIMIZE_ADDRESS): Define.
17002 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
17003 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
17004 (TARGET_ICT_MODEL_SMALL): Define.
17005 (TARGET_ICT_MODEL_LARGE): Define.
17006 * config/nds32/nds32.md (movsi): Consider ict model.
17007 (call, call_value): Consider ict model.
17008 (sibcall, sibcall_value): Consider ict model.
17009 * config/nds32/nds32.opt (mict-model): New option.
17010 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
17013 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
17014 Monk Chiang <sh.chiang04@gmail.com>
17015 Jim Wilson <jimw@sifive.com>
17017 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
17018 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
17019 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
17020 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
17021 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
17022 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
17023 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
17024 compute save_libcall_adjustment properly.
17025 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
17026 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
17027 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
17028 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
17029 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
17030 (ABI_SPEC): Handle mabi=ilp32e.
17031 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
17032 (RVE): Add RVE mask.
17033 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
17034 <-march>: Add rv32e as an example.
17036 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
17039 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
17040 (intra_create_variable_infos): Handle C++ constructors.
17042 2018-05-18 Martin Liska <mliska@suse.cz>
17044 * passes.def: Remove a redundant pass.
17046 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
17049 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
17051 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17053 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
17055 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
17057 (armv2, armv2a, armv3, armv3m): Delete architectures.
17058 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
17059 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
17060 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
17062 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
17063 (*mulsidi3adddi): Likewise.
17064 (mulsidi3): Likewise.
17065 (*mulsidi3_nov6): Likewise.
17066 (umulsidi3): Likewise.
17067 (umulsidi3_nov6): Likewise.
17068 (umaddsidi4): Likewise.
17069 (*umulsidi3adddi): Likewise.
17070 (smulsi3_highpart): Likewise.
17071 (*smulsi3_highpart_nov6): Likewise.
17072 (umulsi3_highpart): Likewise.
17073 (*umulsi3_highpart_nov6): Likewise.
17074 * config/arm/arm.h (arm_arch3m): Delete.
17075 * config/arm/arm.c (arm_arch3m): Delete.
17076 (arm_option_override_internal): Update armv3-related comment.
17077 (arm_configure_build_target): Delete use of isa_bit_mode32.
17078 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
17079 (arm_rtx_costs_internal): Delete check of arm_arch3m.
17080 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
17081 (mulsa3): Likewise.
17082 (mulusa3): Likewise.
17083 * config/arm/arm-protos.h (arm_arch3m): Delete.
17084 * config/arm/arm-tables.opt: Regenerate.
17085 * config/arm/arm-tune.md: Likewise.
17086 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
17087 deleted architectures.
17089 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17091 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
17092 (armv5t, armv5te): New features.
17093 (ARMv5, ARMv5e): Delete fgroups.
17094 (ARMv5t, ARMv5te): Adjust for above changes.
17095 (ARMv6m): Likewise.
17096 (armv5, armv5e): Delete arches.
17097 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
17099 (*call_reg_arm): Likewise.
17100 (*call_value_reg_armv5): Likewise.
17101 (*call_value_reg_arm): Likewise.
17102 (*call_symbol): Likewise.
17103 (*call_value_symbol): Likewise.
17104 (*sibcall_insn): Likewise.
17105 (*sibcall_value_insn): Likewise.
17106 (clzsi2): Likewise.
17107 (prefetch): Likewise.
17108 (define_split and define_peephole2 dependent on arm_arch5):
17110 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
17112 (TARGET_ARM_QBIT): Likewise.
17113 (TARGET_DSP_MULTIPLY): Likewise.
17114 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
17115 (arm_arch5, arm_arch5e): Delete.
17116 (arm_arch5t, arm_arch5te): Declare.
17117 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
17118 (arm_arch5t): Declare.
17119 (arm_option_reconfigure_globals): Update for the above.
17120 (arm_options_perform_arch_sanity_checks): Update comment, replace
17121 use of arm_arch5 with arm_arch5t.
17122 (use_return_insn): Likewise.
17123 (arm_emit_call_insn): Likewise.
17124 (output_return_instruction): Likewise.
17125 (arm_final_prescan_insn): Likewise.
17126 (arm_coproc_builtin_available): Likewise.
17127 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
17128 arm_arch5e with arm_arch5t and arm_arch5te.
17129 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
17130 (arm_arch5t, arm_arch5te): Declare.
17131 * config/arm/arm-tables.opt: Regenerate.
17132 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
17133 * config/arm/t-multilib: Likewise.
17134 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
17135 instead of arm_arch5.
17136 (*call_reg_thumb1): Likewise.
17137 (*call_value_reg_thumb1_v5): Likewise.
17138 (*call_value_reg_thumb1): Likewise.
17139 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
17141 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
17143 2018-05-18 Martin Liska <mliska@suse.cz>
17145 PR gcov-profile/84846
17146 * doc/gcov.texi: Document -t option of gcov tool.
17148 2018-05-18 Martin Liska <mliska@suse.cz>
17150 PR gcov-profile/84846
17151 * gcov.c (print_usage): Add new -t option.
17152 (process_args): Handle the option.
17153 (generate_results): Use stdout as output when requested by
17156 2018-05-18 Martin Liska <mliska@suse.cz>
17158 PR gcov-profile/84846
17159 * coverage.c (coverage_init): Write PWD to .gcno file.
17160 * doc/gcov.texi: Document how working directory is printed.
17161 * gcov-dump.c (dump_gcov_file): Print PWD.
17162 * gcov.c (output_intermediate_file): Likewise.
17163 (read_graph_file): Read PWD string.
17164 (output_lines): Print PWD.
17166 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17168 PR middle-end/85817
17169 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
17170 for retval and return false if all args to phi are zero.
17172 2018-05-18 Richard Biener <rguenther@suse.de>
17174 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
17176 (evrp_dom_walker::before_dom_children): Call it.
17178 2018-05-18 Richard Biener <rguenther@suse.de>
17180 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
17181 results when processing array refs with variable index.
17183 2018-05-18 Toon Moene <toon@moene.org>
17185 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
17186 directly after that of -floop-interchange. Indicate that both
17187 options are enabled by default when specifying -O3.
17189 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17191 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
17192 iterator. Delete separate integer-mode vec_set<mode> expander.
17193 (aarch64_simd_vec_setv2di): Delete.
17194 (vec_setv2di): Delete.
17195 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
17196 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
17197 the "w, r" alternative.
17199 2018-05-18 Martin Liska <mliska@suse.cz>
17201 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
17202 * tree-pass.h (make_pass_lower_switch_O0): New function.
17203 * tree-switch-conversion.c (node_has_low_bound): Remove.
17204 (node_has_high_bound): Likewise.
17205 (node_is_bounded): Likewise.
17206 (class pass_lower_switch): Make it a template type and create
17208 (pass_lower_switch::execute): Add template argument.
17209 (make_pass_lower_switch): New function.
17210 (make_pass_lower_switch_O0): New function.
17211 (do_jump_if_equal): Remove.
17212 (emit_case_nodes): Simplify to just handle all 3 cases and leave
17213 all the hard work to tree optimization passes.
17215 2018-05-18 Martin Liska <mliska@suse.cz>
17217 * dbgcnt.c (limit_low): Renamed from limit.
17218 (limit_high): New variable.
17219 (dbg_cnt_is_enabled): Check for upper limit.
17220 (dbg_cnt): Adjust dumping.
17221 (dbg_cnt_set_limit_by_index): Add new argument for high
17223 (dbg_cnt_set_limit_by_name): Likewise.
17224 (dbg_cnt_process_single_pair): Parse new format.
17225 (dbg_cnt_process_opt): Use strtok.
17226 (dbg_cnt_list_all_counters): Remove 'value' and add
17228 * doc/invoke.texi: Document changes.
17230 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
17232 * doc/sourcebuild.texi (scalar_all_fma): Document.
17233 * tree.def (FMA_EXPR): Delete.
17234 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
17235 * internal-fn.c (ternary_direct): New macro.
17236 (expand_ternary_optab_fn): Likewise.
17237 (direct_ternary_optab_supported_p): Likewise.
17238 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
17239 * builtins.c (fold_builtin_fma): Delete.
17240 (fold_builtin_3): Don't call it.
17241 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
17242 * expr.c (expand_expr_real_2): Likewise.
17243 * fold-const.c (operand_equal_p): Likewise.
17244 (fold_ternary_loc): Likewise.
17245 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
17246 * gimple.c (DEFTREECODE): Likewise.
17247 * gimplify.c (gimplify_expr): Likewise.
17248 * optabs-tree.c (optab_for_tree_code): Likewise.
17249 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17250 * tree-eh.c (operation_could_trap_p): Likewise.
17251 (stmt_could_throw_1_p): Likewise.
17252 * tree-inline.c (estimate_operator_cost): Likewise.
17253 * tree-pretty-print.c (dump_generic_node): Likewise.
17254 (op_code_prio): Likewise.
17255 * tree-ssa-loop-im.c (stmt_cost): Likewise.
17256 * tree-ssa-operands.c (get_expr_operands): Likewise.
17257 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
17258 * fold-const-call.h (fold_fma): Delete.
17259 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
17260 CFN_FNMA and CFN_FNMS.
17261 (fold_fma): Delete.
17262 * genmatch.c (combined_fn): New enum.
17263 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
17264 (commutative_op): New function.
17265 (commutate): Use it. Handle more than 2 operands.
17266 (dt_operand::gen_gimple_expr): Use commutative_op.
17267 (parser::parse_expr): Allow :c to be used with non-binary
17268 operators if the commutative operand is known.
17269 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
17270 CFN_FMS, CFN_FNMA and CFN_FNMS.
17271 (backprop::process_assign_use): Remove FMA_EXPR handling.
17272 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
17273 (gen_hsa_fma): New function.
17274 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
17275 IFN_FNMA and IFN_FNMS.
17276 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
17277 * gimple-fold.h (follow_all_ssa_edges): Declare.
17278 * gimple-fold.c (follow_all_ssa_edges): New function.
17279 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
17280 gimple_build interface and use follow_all_ssa_edges to fold the result.
17281 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
17282 instead of checking for optabs directly.
17283 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
17284 rather than FMA_EXPRs.
17285 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
17286 call to IFN_FMA instead of an FMA_EXPR.
17288 2018-05-17 Jim Wilson <jimw@sifive.com>
17290 * expr.c (do_tablejump): When converting index to Pmode, if we have a
17291 sign extended promoted subreg, and the range does not have the sign bit
17292 set, then do a sign extend.
17294 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
17295 test, check for sign extended subreg and/or constant operands, and
17296 do a sign extend in that case.
17298 2018-05-17 Steve Ellcey <sellcey@cavium.com>
17300 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
17301 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
17303 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
17304 Change logics_shift_reg to logics_shift_imm.
17305 (thunderx2t99_fp_loadpair_basic): Delete.
17306 (thunderx2t99_fp_storepair_basic): Delete.
17307 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
17308 (thunderx2t99_asimd_polynomial): Delete.
17309 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
17310 and neon_fp_mul_d_scalar_q.
17311 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
17312 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
17313 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
17314 (thunderx2t99_asimd_lut): Add missing tbl types.
17315 (thunderx2t99_asimd_ext): Delete.
17316 (thunderx2t99_asimd_load1_1_mult): Delete.
17317 (thunderx2t99_asimd_load1_2_mult): Delete.
17318 (thunderx2t99_asimd_load1_ldp): New.
17319 (thunderx2t99_asimd_load1): New.
17320 (thunderx2t99_asimd_load2): Add missing *load2* types.
17321 (thunderx2t99_asimd_load3): New.
17322 (thunderx2t99_asimd_load4): New.
17323 (thunderx2t99_asimd_store1_1_mult): Delete.
17324 (thunderx2t99_asimd_store1_2_mult): Delete.
17325 (thunderx2t99_asimd_store2_mult): Delete.
17326 (thunderx2t99_asimd_store2_onelane): Delete.
17327 (thunderx2t99_asimd_store_stp): New.
17328 (thunderx2t99_asimd_store1): New.
17329 (thunderx2t99_asimd_store2): New.
17330 (thunderx2t99_asimd_store3): New.
17331 (thunderx2t99_asimd_store4): New.
17333 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
17335 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
17336 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
17338 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
17339 Segher Boessenkool <segher@kernel.crashing.org>
17342 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
17345 2018-05-17 Richard Biener <rguenther@suse.de>
17347 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
17348 for pruning loop and prune defs feeding only already visited PHIs.
17350 2018-05-17 Richard Biener <rguenther@suse.de>
17352 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
17354 2018-05-17 Bin Cheng <bin.cheng@arm.com>
17355 Richard Biener <rguenther@suse.de>
17357 PR tree-optimization/85793
17358 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
17359 for VMAT_ELEMENTWISE.
17361 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
17363 * internal-fn.h (lookup_internal_fn): Declare
17364 * internal-fn.c (lookup_internal_fn): New function.
17365 * gimple.c (gimple_build_call_from_tree): Handle calls to
17366 internal functions.
17367 * gimple-pretty-print.c (dump_gimple_call): Print "." before
17368 internal function names.
17369 * tree-pretty-print.c (dump_generic_node): Likewise.
17370 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
17372 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
17374 * gimple-fold.h (gimple_build): Make the function forms take
17375 combined_fn rather than built_in_function.
17376 (gimple_simplify): Likewise.
17377 * gimple-match-head.c (gimple_simplify): Likewise.
17378 * gimple-fold.c (gimple_build): Likewise.
17379 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
17380 rather than gimple_build_call_internal.
17381 (get_initial_defs_for_reduction): Likewise.
17382 (vect_create_epilog_for_reduction): Likewise.
17383 (vectorizable_live_operation): Likewise.
17385 2018-05-17 Martin Liska <mliska@suse.cz>
17387 * gimple-ssa-sprintf.c (format_directive): Do not use
17388 space in between 'G_' and '('.
17390 2018-05-17 Jakub Jelinek <jakub@redhat.com>
17393 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
17394 even if the mask is not all ones.
17397 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
17399 (ix86_gimple_fold_builtin): Likewise.
17402 * config/i386/i386.c: Include tree-vector-builder.h.
17403 (ix86_vector_shift_count): New function.
17404 (ix86_fold_builtin): Fold shift builtins by scalar count.
17405 (ix86_gimple_fold_builtin): Likewise.
17407 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
17408 _mm512_setzero): New intrinsics.
17410 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
17411 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17413 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
17414 code generation for cases where splatting a value is not useful.
17415 * simplify-rtx.c (simplify_ternary_operation): Simplify
17416 vec_merge across a vec_duplicate and a paradoxical subreg forming
17417 a vector mode to a vec_concat.
17419 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
17421 * config.gcc: Support "goldmont-plus".
17422 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
17424 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17425 PROCESSOR_GOLDMONT_PLUS.
17426 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
17427 (processor_target_table): Add "goldmont-plus".
17428 (PTA_GOLDMONT_PLUS): Define.
17429 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
17430 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
17431 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
17432 (fold_builtin_cpu): Add "goldmont-plus".
17433 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
17434 (ix86_option_override_internal): Add "goldmont-plus".
17435 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
17436 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
17437 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
17438 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
17440 2018-05-17 Richard Biener <rguenther@suse.de>
17442 PR tree-optimization/85757
17443 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
17444 remove defs that only feed that PHI from further processing.
17446 2018-05-16 Jim Wilson <jimw@sifive.com>
17448 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
17450 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
17452 2018-05-16 Mark Wielaard <mark@klomp.org>
17454 * dwarf2out.c (count_index_strings): New function.
17455 (output_indirect_strings): Call count_index_strings and generate
17456 header for dwarf_version >= 5.
17458 2018-05-16 Mark Wielaard <mark@klomp.org>
17460 * dwarf2out.c (dwarf_FORM): New function.
17461 (set_indirect_string): Use dwarf_FORM.
17462 (reset_indirect_string): Likewise.
17463 (size_of_die): Likewise.
17464 (value_format): Likewise.
17465 (output_die): Likewise.
17466 (add_skeleton_AT_string): Likewise.
17467 (output_macinfo_op): Likewise.
17468 (index_string): Likewise.
17469 (output_index_string_offset): Likewise.
17470 (output_index_string): Likewise.
17471 (count_index_strings): Likewise.
17473 2018-05-16 Carl Love <cel@us.ibm.com>
17475 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
17476 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
17478 2018-05-16 Martin Jambor <mjambor@suse.cz>
17480 * ipa-prop.c (ipa_free_all_edge_args): Remove.
17481 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
17483 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
17485 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
17486 (fnma<mode>4): Likewise.
17487 (fms<mode>4): Likewise.
17488 (fnms<mode>4): Likewise.
17489 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
17490 (aarch64_fnma<mode>4): Likewise.
17491 (aarch64_fms<mode>4): Likewise.
17492 (aarch64_fnms<mode>4): Likewise.
17493 (aarch64_fnmadd<mode>4): Likewise.
17495 2018-05-16 Jason Merrill <jason@redhat.com>
17497 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
17499 2018-05-16 Richard Biener <rguenther@suse.de>
17501 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
17502 (dump_stmt_cost): Declare.
17503 (add_stmt_cost): Dump cost we add.
17504 (add_stmt_costs): New function.
17505 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
17506 No longer exported.
17507 (vect_analyze_stmt): Adjust prototype.
17508 (vectorizable_condition): Likewise.
17509 (vectorizable_live_operation): Likewise.
17510 (vectorizable_reduction): Likewise.
17511 (vectorizable_induction): Likewise.
17512 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
17513 cost vector to pass to vectorizable_ and record afterwards.
17514 (vect_model_reduction_cost): Take cost vector argument and adjust.
17515 (vect_model_induction_cost): Likewise.
17516 (vectorizable_reduction): Likewise.
17517 (vectorizable_induction): Likewise.
17518 (vectorizable_live_operation): Likewise.
17519 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
17520 SLP_TREE_NUMBER_OF_VEC_STMTS.
17521 (vect_analyze_slp_cost_1): Remove.
17522 (vect_analyze_slp_cost): Likewise.
17523 (vect_slp_analyze_node_operations): Take visited args and
17524 a target cost vector. Avoid processing already visited stmt sets.
17525 (vect_slp_analyze_operations): Use a local cost vector to gather
17526 costs and register those of non-discarded instances.
17527 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
17528 (vect_schedule_slp_instance): Remove copying of
17529 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
17531 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
17532 adding cost. Record cost entry location.
17533 (vect_prologue_cost_for_slp_op): Function to compute cost of
17534 a constant or invariant generated for SLP vect in the prologue,
17535 split out from vect_analyze_slp_cost_1.
17536 (vect_model_simple_cost): Make static. Adjust for SLP costing.
17537 (vect_model_promotion_demotion_cost): Likewise.
17538 (vect_model_store_cost): Likewise, make static.
17539 (vect_model_load_cost): Likewise.
17540 (vectorizable_bswap): Add cost vector arg and adjust.
17541 (vectorizable_call): Likewise.
17542 (vectorizable_simd_clone_call): Likewise.
17543 (vectorizable_conversion): Likewise.
17544 (vectorizable_assignment): Likewise.
17545 (vectorizable_shift): Likewise.
17546 (vectorizable_operation): Likewise.
17547 (vectorizable_store): Likewise.
17548 (vectorizable_load): Likewise.
17549 (vectorizable_condition): Likewise.
17550 (vectorizable_comparison): Likewise.
17551 (can_vectorize_live_stmts): Likewise.
17552 (vect_analyze_stmt): Likewise.
17553 (vect_transform_stmt): Adjust calls to vectorizable_*.
17554 * tree-vectorizer.c: Include gimple-pretty-print.h.
17555 (dump_stmt_cost): New function.
17557 2018-05-16 Richard Biener <rguenther@suse.de>
17559 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
17560 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
17561 * tree-ssa-dse.c: Include tree-ssa-loop.h.
17562 (check_name): New callback.
17563 (dse_classify_store): Track cycles via a visited bitmap of PHI
17564 defs and simplify handling of in-loop and across loop dead stores
17565 and properly fail for loop-variant refs. Handle byte-tracking with
17566 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
17569 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
17571 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
17572 (vect_get_mask_type_for_stmt): Likewise.
17573 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
17575 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
17576 to determine the statement's vector type and the vector type that
17577 should be used for calculating nunits. Deal with cases in which
17578 the type has to be deferred.
17579 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
17580 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
17581 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
17582 (vect_determine_vf_for_stmt): New functions, split out from...
17583 (vect_determine_vectorization_factor): ...here.
17584 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
17585 (vect_get_mask_type_for_stmt): New functions, split out from
17586 vect_determine_vectorization_factor.
17588 2018-05-16 Richard Biener <rguenther@suse.de>
17590 * tree-cfg.c (verify_gimple_assign_ternary): Properly
17591 verify the [VEC_]COND_EXPR embedded comparison.
17593 2018-05-15 Martin Sebor <msebor@redhat.com>
17595 PR tree-optimization/85753
17596 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
17597 RECORD_TYPE in addition to ARRAY_TYPE.
17599 2018-05-15 Martin Sebor <msebor@redhat.com>
17601 PR middle-end/85643
17602 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
17604 2018-05-15 Richard Biener <rguenther@suse.de>
17606 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
17607 add by_clobber_p one. Change algorithm to collect all defs
17608 representing uses we need to walk and try reducing them to
17609 a single one before failing.
17610 (dse_dom_walker::dse_optimize_stmt): Adjust.
17612 2018-05-13 Mark Wielaard <mark@klomp.org>
17614 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
17615 (size_of_loc_descr): Likewise.
17616 (output_loc_operands): Likewise.
17617 (output_loc_operands_raw): Likewise.
17618 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
17619 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
17620 (hash_loc_operands): Likewise.
17621 (compare_loc_operands): Likewise.
17623 2018-05-14 Mark Wielaard <mark@klomp.org>
17625 * dwarf2out.c (count_index_addrs): New function.
17626 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
17628 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17630 PR tree-optimization/83648
17631 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
17632 return value as malloc candidate.
17634 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17637 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
17638 param as true in call to suggest_attribute.
17640 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
17642 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
17643 -mreadonly-in-sdata.
17645 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17647 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
17649 (aarch64_crypto_aesd_fused): Likewise.
17651 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
17653 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
17654 (movsi_aarch64): Likewise.
17655 (load_pairsi): Likewise.
17656 (load_pairdi): Likewise.
17657 (store_pairsi): Likewise.
17658 (store_pairdi): Likewise.
17659 (load_pairsf): Likewise.
17660 (load_pairdf): Likewise.
17661 (store_pairsf): Likewise.
17662 (store_pairdf): Likewise.
17663 (zero_extend): Likewise.
17664 (trunc): Swap alternatives.
17665 (fcvt_target): Add '?' to prefer w over r.
17667 2018-05-14 Jakub Jelinek <jakub@redhat.com>
17670 * config/i386/i386.md: Disallow non-commutative arithmetics in
17671 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
17672 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
17673 in the peephole2 before it.
17675 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
17677 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
17678 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
17679 (ix86_handle_option): Handle -mcldemote.
17680 * config.gcc: New header.
17681 * config/i386/cldemoteintrin.h: New file.
17682 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
17683 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
17685 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17686 OPTION_MASK_ISA_CLDEMOTE.
17687 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
17688 (ix86_valid_target_attribute_inner_p): Ditto.
17689 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
17690 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
17691 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
17692 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
17693 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
17695 * config/i386/i386.opt: Add -mcldemote.
17696 * config/i386/x86intrin.h: New header.
17697 * doc/invoke.texi: Add -mcldemote.
17699 2018-05-14 Richard Biener <rguenther@suse.de>
17701 * doc/match-and-simplify.texi: Adjust :s documentation.
17703 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
17705 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
17706 intended memcpy size.
17707 (REORDER_45): Likewise.
17709 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
17711 * sort.cc: New file.
17712 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
17713 * vec.c (qsort_chk): Use gcc_qsort.
17714 * Makefile.in (OBJS-libcommon): Add sort.o.
17715 (build/sort.o): New target. Use it...
17716 (BUILD_RTL): ... here, and...
17717 (build/gencfn-macros): ... here, and...
17718 (build/genmatch): ... here.
17720 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
17721 Chung-Ju Wu <jasonwucj@gmail.com>
17723 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
17724 * config/nds32/nds32-graywolf.md: New file.
17725 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
17726 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
17728 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
17729 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
17730 * config/nds32/nds32.md (pipeline_model): Add graywolf.
17731 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
17732 * config/nds32/pipelines.md: Include n15 settings.
17734 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
17735 Chung-Ju Wu <jasonwucj@gmail.com>
17737 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
17738 * config/nds32/nds32-n13.md: New file.
17739 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
17740 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
17742 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
17743 * config/nds32/nds32.md (pipeline_model): Add n13.
17744 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
17745 * config/nds32/pipelines.md: Include n13 settings.
17747 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
17748 Chung-Ju Wu <jasonwucj@gmail.com>
17750 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
17751 * config/nds32/nds32-n10.md: New file.
17752 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
17753 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
17755 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
17756 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
17757 * config/nds32/nds32.md (pipeline_model): Add n10.
17758 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
17759 * config/nds32/pipelines.md: Include n10 settings.
17761 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
17762 Kito Cheng <kito.cheng@gmail.com>
17763 Chung-Ju Wu <jasonwucj@gmail.com>
17765 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
17766 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
17767 Add enum values for DSP extension instructions.
17768 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
17770 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
17771 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
17772 New code iterators.
17773 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
17774 * config/nds32/nds32-dspext.md: New file for DSP implementation.
17775 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
17776 * config/nds32/nds32-intrinsic.md: Likewise.
17777 * config/nds32/nds32_intrinsic.h: Likewise.
17778 * config/nds32/nds32-md-auxiliary.c: Likewise.
17779 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
17780 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
17781 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
17782 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
17783 * config/nds32/nds32-protos.h: New declarations for DSP extension.
17784 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
17785 TYPE_DMAC in switch statement.
17786 * config/nds32/nds32.c: New checking and implementation for DSP
17787 extension instructions.
17788 * config/nds32/nds32.h: Likewise.
17789 * config/nds32/nds32.md: Likewise.
17790 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
17791 * config/nds32/predicates.md: Implement new predicates for DSP
17794 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
17796 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
17797 Reformat alternatives and attributes so it is easier to identify
17798 which constraints/attributes go with which instruction.
17799 (mov<mode>_hardfloat32, FMOVE64): Likewise.
17800 (mov<mode>_softfloat32, FMOVE64): Likewise.
17801 (mov<mode>_hardfloat64, FMOVE64): Likewise.
17802 (mov<mode>_softfloat64, FMOVE64): Likewise.
17804 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
17806 * doc/extend.texi (PowerPC Built-in Functions): Rename this
17808 (Basic PowerPC Built-in Functions): The new name of the
17809 subsection previously known as "PowerPC Built-in Functions".
17810 (Basic PowerPC Built-in Functions Available on all Configurations):
17812 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
17813 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
17814 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
17815 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
17817 2018-05-11 Martin Jambor <mjambor@suse.cz>
17820 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
17823 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
17826 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
17828 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
17830 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
17831 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
17832 (ix86_handle_option): Handle -mwaitpkg.
17833 * config.gcc: New header.
17834 * config/i386/cpuid.h (bit_WAITPKG): New bit.
17835 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
17836 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
17838 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17839 OPTION_MASK_ISA_WAITPKG.
17840 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
17841 (ix86_option_override_internal): Add PTA_WAITPKG.
17842 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
17843 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
17844 IX86_BUILTIN_TPAUSE.
17845 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
17846 __builtin_ia32_umwait and __builtin_ia32_tpause.
17847 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
17848 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
17849 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
17850 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
17851 UNSPECV_TPAUSE): New.
17852 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
17853 * config/i386/i386.opt: Add -mwaitpkg.
17854 * config/i386/waitpkgintrin.h: New file.
17855 * config/i386/x86intrin.h: New header.
17856 * doc/invoke.texi: Add -mwaitpkg.
17858 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
17861 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
17863 (cortex-m0): Use armv6s-m isa.
17864 (cortex-m0plus): Likewise.
17865 (cortex-m1): Likewise.
17866 (cortex-m0.small-multiply): Likewise.
17867 (cortex-m0plus.small-multiply): Likewise.
17868 (cortex-m1.small-multiply): Likewise.
17870 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
17871 Jakub Jelinek <jakub@redhat.com>
17873 PR tree-optimization/85692
17874 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
17875 source permute as well.
17877 2018-05-11 Martin Liska <mliska@suse.cz>
17880 * doc/extend.texi: Document LLVM style format for no_sanitize
17883 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
17885 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
17886 mode_supports_vsx_dform_quad to mode_supports_dq_form.
17887 (mode_supports_vsx_dform_quad): Likewise.
17888 (mode_supports_vmx_dform): Move these functions to be next to the
17889 other mode_supports functions.
17890 (mode_supports_dq_form): Likewise.
17891 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
17892 mode_supports_dq_form.
17893 (reg_offset_addressing_ok_p): Likewise.
17894 (offsettable_ok_by_alignment): Likewise.
17895 (rs6000_legitimate_offset_address_p): Likewise.
17896 (legitimate_lo_sum_address_p): Likewise.
17897 (rs6000_legitimize_address): Likewise.
17898 (rs6000_legitimize_reload_address): Likewise.
17899 (rs6000_secondary_reload_inner): Likewise.
17900 (rs6000_preferred_reload_class): Likewise.
17901 (rs6000_output_move_128bit): Likewise.
17903 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
17905 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
17906 Generate SImode target register for null target.
17907 <case IX86_BUILTIN_XGETBV>: Ditto.
17908 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
17909 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
17911 2018-05-10 Carl Love <cel@us.ibm.com>
17913 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
17914 dcbtt and dcbtstt if operands[2] is 0.
17916 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
17919 * config/i386/sse.md (usadv64qi): New expander.
17921 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
17923 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
17924 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
17925 -maltivec=be support.
17926 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
17927 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
17928 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17929 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
17930 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
17931 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
17932 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
17933 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
17934 altivec_vsumsws): Adjust.
17935 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
17936 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
17937 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
17939 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
17940 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
17941 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
17942 (altivec_lve<VI_char>x): Delete expand.
17943 (*altivec_lve<VI_char>x_internal): Rename to...
17944 (altivec_lve<VI_char>x): ... this.
17945 (altivec_lvxl_<mode>): Delete expand.
17946 (*altivec_lvxl_<mode>_internal): Rename to ...
17947 (altivec_lvxl_<mode>): ... this.
17948 (altivec_stvxl_<mode>): Delete expand.
17949 (*altivec_stvxl_<mode>_internal): Rename to ...
17950 (altivec_stvxl_<mode>): ... this.
17951 (altivec_stve<VI_char>x): Delete expand.
17952 (*altivec_stve<VI_char>x_internal): Rename to ...
17953 (altivec_stve<VI_char>x): ... this.
17954 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
17955 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
17956 reduc_plus_scal_<mode>): Adjust.
17957 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
17959 (rs6000_cpu_cpp_builtins): Adjust.
17960 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
17961 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
17962 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
17963 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
17964 -maltivec=be support.
17965 (rs6000_split_vec_extract_var): Adjust.
17966 (rs6000_split_v4si_init): Adjust.
17967 (swap_selector_for_mode): Delete.
17968 (altivec_expand_lvx_be, altivec_expand_stvx_be,
17969 altivec_expand_stvex_be): Delete.
17970 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
17971 -maltivec=be support.
17972 (rs6000_gimple_fold_builtin): Ditto.
17973 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
17975 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
17976 (TARGET_DIRECT_MOVE_64BIT): Adjust.
17977 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
17978 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
17979 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
17980 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
17981 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
17982 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
17983 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
17984 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
17985 anonymous split): Adjust.
17986 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
17987 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
17989 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
17991 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
17992 when --with-gxx-include-dir is also specified.
17993 * configure: Regenerate.
17995 2018-05-09 Jim Wilson <jimw@sifive.com>
17998 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
17999 * config/riscv/t-withmultilib: New.
18000 * config/riscv/withmultilib.h: New.
18001 * doc/install.texi: Document RISC-V --with-multilib-list support.
18003 2018-05-09 Richard Biener <rguenther@suse.de>
18005 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
18007 (vect_bb_vectorization_profitable_p): Adjust. Compute
18008 actual scalar cost using the cost vector and the add_stmt_cost
18011 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
18013 PR rtl-optimization/85645
18014 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
18015 in the REG_CFA_REGISTER note for LR, don't leave it empty.
18017 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
18019 PR rtl-optimization/85645
18020 * shrink-wrap.c (spread_components): Return a boolean saying if
18021 anything was changed.
18022 (try_shrink_wrapping_separate): Iterate spread_components until
18023 nothing changes anymore.
18025 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
18027 PR rtl-optimization/85645
18028 * regrename.c (build_def_use): Also kill the chains that include the
18029 destination of a REG_CFA_REGISTER note.
18031 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
18033 PR rtl-optimization/85645
18034 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
18035 insn that has a REG_CFA_REGISTER note.
18037 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
18039 * cfgexpand.c (expand_clobber): New function.
18040 (expand_gimple_stmt_1): Use it.
18041 * tree-vect-stmts.c (vect_clobber_variable): New function,
18043 (vectorizable_simd_clone_call): ...here.
18044 (vectorizable_store): Emit a clobber either side of an
18045 IFN_STORE_LANES sequence.
18046 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
18048 2018-05-09 Tom de Vries <tom@codesourcery.com>
18051 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
18052 (define_insn "trap_if_false"): Add exit after trap.
18054 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
18056 PR rtl-optimization/85638
18057 * bb-reorder.c: Include common/common-target.h.
18058 (create_forwarder_block): New function extracted from...
18059 (fix_up_crossing_landing_pad): ...here. Rename into...
18060 (dw2_fix_up_crossing_landing_pad): ...this.
18061 (sjlj_fix_up_crossing_landing_pad): New function.
18062 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
18063 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
18064 from both partitions and exit the loop after one iteration.
18066 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
18069 * doc/extend.texi (PowerPC Built-in Functions): Rename this
18071 (Basic PowerPC Built-in Functions): The new name of the
18072 subsection previously known as "PowerPC Built-in Functions".
18073 (Basic PowerPC Built-in Functions Available on all Configurations):
18075 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
18077 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
18079 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
18081 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
18084 2018-05-08 Jim Wilson <jimw@sifive.com>
18086 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
18087 (LD_EMUL_SUFFIX): New.
18088 (LINK_SPEC): Use it.
18090 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
18092 * doc/extend.texi (PowerPC Built-in Functions): Rename this
18094 (Basic PowerPC Built-in Functions): The new name of the
18095 subsection previously known as "PowerPC Built-in Functions".
18096 (Basic PowerPC Built-in Functions Available on all Configurations):
18098 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
18100 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
18102 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
18104 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
18107 2018-05-08 Jakub Jelinek <jakub@redhat.com>
18110 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
18111 after cmpelim optimization.
18113 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
18115 * config.gcc: Support "goldmont".
18116 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
18117 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
18118 PROCESSOR_GOLDMONT.
18119 * config/i386/i386.c (m_GOLDMONT): Define.
18120 (processor_target_table): Add "goldmont".
18121 (PTA_GOLDMONT): Define.
18122 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
18123 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
18124 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
18125 (fold_builtin_cpu): Add "goldmont".
18126 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
18127 (ix86_option_override_internal): Add "goldmont".
18128 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
18129 (processor_type): Add PROCESSOR_GOLDMONT.
18130 * config/i386/i386.md: Add CPU "glm".
18131 * config/i386/glm.md: New file.
18132 * config/i386/x86-tune.def: Add m_GOLDMONT.
18133 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
18135 2018-05-08 Jakub Jelinek <jakub@redhat.com>
18138 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
18140 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
18141 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
18142 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
18145 * config/i386/i386.c (ix86_fold_builtin): Handle
18146 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
18149 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
18150 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
18152 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
18155 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
18156 (check_arch): Likewise.
18157 (check_fpu): Return the result rather than printing it.
18158 (end arch): Fix operator precedence.
18159 (end cpu): Likewise.
18160 (END): Print the result from check_fpu.
18162 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
18163 Alan Hayward <alan.hayward@arm.com>
18164 David Sherwood <david.sherwood@arm.com>
18166 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
18167 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
18168 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
18169 (*fcmuo<mode>_and): New patterns.
18171 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
18173 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
18174 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
18175 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
18176 (cmp_op, sve_imm_con): New code attributes.
18177 (SVE_COND_INT_CMP, imm_con): Delete.
18178 (cmp_op): Remove above unspecs from int attribute.
18179 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
18181 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
18182 comparison-specific unspecs.
18183 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
18184 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
18185 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
18186 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
18187 (*vec_fcm<cmp_op><mode>): Rename to...
18188 (*fcm<cmp_op><mode>): ...this and adjust likewise.
18189 (*vec_fcmuo<mode>): Rename to...
18190 (*fcmuo<mode>): ...this and adjust likewise.
18191 (*pred_fcm<cmp_op><mode>): New pattern.
18192 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
18193 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
18195 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
18197 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
18198 (aarch64_emit_sve_predicated_cond): New function.
18199 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
18200 (aarch64_emit_unspec_cond_or): Replace with...
18201 (aarch64_emit_sve_or_conds): ...this new function. Use
18202 aarch64_emit_sve_ptrue_op for the individual comparisons and
18203 aarch64_emit_binop to OR them together.
18204 (aarch64_emit_inverted_unspec_cond): Replace with...
18205 (aarch64_emit_sve_inverted_cond): ...this new function. Use
18206 aarch64_emit_sve_ptrue_op for the comparison and
18207 aarch64_emit_unop to invert the result.
18208 (aarch64_expand_sve_vec_cmp_float): Update after the above
18209 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
18211 2018-05-07 Nathan Sidwell <nathan@acm.org>
18213 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
18214 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
18215 (Backwards Compatibility): Likewise.
18217 2018-05-07 Luis Machado <luis.machado@linaro.org>
18221 2018-05-07 Luis Machado <luis.machado@linaro.org>
18223 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
18224 <prefetch_dynamic_strides>: New const bool field.
18225 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
18226 prefetch_dynamic_strides.
18227 (exynosm1_prefetch_tune): Likewise.
18228 (thunderxt88_prefetch_tune): Likewise.
18229 (thunderx_prefetch_tune): Likewise.
18230 (thunderx2t99_prefetch_tune): Likewise.
18231 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
18233 (aarch64_override_options_internal): Update to set
18234 PARAM_PREFETCH_DYNAMIC_STRIDES.
18235 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
18236 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
18237 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
18238 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
18239 prefetch-dynamic-strides setting.
18241 2018-05-07 Luis Machado <luis.machado@linaro.org>
18243 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
18244 <minimum_stride>: New const int field.
18245 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
18246 minimum_stride field.
18247 (exynosm1_prefetch_tune): Likewise.
18248 (thunderxt88_prefetch_tune): Likewise.
18249 (thunderx_prefetch_tune): Likewise.
18250 (thunderx2t99_prefetch_tune): Likewise.
18251 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
18252 (aarch64_override_options_internal): Update to set
18253 PARAM_PREFETCH_MINIMUM_STRIDE.
18254 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
18255 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
18256 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
18257 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
18258 stride is constant and is below the minimum stride threshold.
18260 2018-05-07 Luis Machado <luis.machado@linaro.org>
18262 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
18265 2018-05-07 Luis Machado <luis.machado@linaro.org>
18267 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
18268 <prefetch_dynamic_strides>: New const bool field.
18269 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
18270 prefetch_dynamic_strides.
18271 (exynosm1_prefetch_tune): Likewise.
18272 (thunderxt88_prefetch_tune): Likewise.
18273 (thunderx_prefetch_tune): Likewise.
18274 (thunderx2t99_prefetch_tune): Likewise.
18275 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
18277 (aarch64_override_options_internal): Update to set
18278 PARAM_PREFETCH_DYNAMIC_STRIDES.
18279 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
18280 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
18281 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
18282 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
18283 prefetch-dynamic-strides setting.
18285 2018-05-07 Luis Machado <luis.machado@linaro.org>
18287 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
18288 <minimum_stride>: New const int field.
18289 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
18290 minimum_stride field.
18291 (exynosm1_prefetch_tune): Likewise.
18292 (thunderxt88_prefetch_tune): Likewise.
18293 (thunderx_prefetch_tune): Likewise.
18294 (thunderx2t99_prefetch_tune): Likewise.
18295 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
18296 (aarch64_override_options_internal): Update to set
18297 PARAM_PREFETCH_MINIMUM_STRIDE.
18298 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
18299 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
18300 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
18301 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
18302 stride is constant and is below the minimum stride threshold.
18304 2018-05-06 Jakub Jelinek <jakub@redhat.com>
18307 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
18308 the type is addressable. Don't force op into register if it has
18311 2018-05-05 Roland McGrath <mcgrathr@google.com>
18314 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
18315 any section for which we don't know a specific type it should have,
18316 regardless of name. Previously this was done only for the exact
18317 names ".init_array", ".fini_array", and ".preinit_array".
18318 (default_elf_asm_named_section): Add comment about
18319 relationship with default_section_type_flags and SECTION_NOTYPE.
18320 (get_section): Don't consider it a type conflict if one side has
18321 SECTION_NOTYPE and the other doesn't, as long as neither has the
18322 SECTION_BSS et al used in the default_section_type_flags logic.
18324 2018-05-05 Tom de Vries <tom@codesourcery.com>
18327 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
18328 (workaround_barsyncs): New function.
18329 (nvptx_reorg): Use workaround_barsyncs.
18330 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
18331 (define_expand "nvptx_membar_cta"): New define_expand.
18332 (define_insn "*nvptx_membar_cta"): New insn.
18334 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
18336 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
18337 To improve optimization opportunities.
18338 * builtin-types.def: The new needed builtin types for the above.
18340 2018-05-04 Richard Biener <rguenther@suse.de>
18342 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
18343 * gimple-ssa-store-merging.c
18344 (imm_store_chain_info::output_merged_store): Remove redundant create,
18345 release split_store vector contents on failure.
18346 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
18347 scalar stmt vector on cache hit.
18349 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
18351 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
18353 * config.gcc (powerpc-xilinx-eabi*): Remove.
18354 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
18356 (fusion_addis_mem_combo_load): Ditto.
18357 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
18359 (rs6000_cpu_cpp_builtins): Ditto.
18360 * config/rs6000/rs6000-linux.c
18361 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
18362 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
18363 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
18365 (rs6000_setup_reg_addr_masks): Ditto.
18366 (rs6000_init_hard_regno_mode_ok): Ditto.
18367 (rs6000_option_override_internal): Ditto.
18368 (legitimate_lo_sum_address_p): Ditto.
18369 (rs6000_legitimize_address): Ditto.
18370 (rs6000_legitimize_reload_address): Ditto.
18371 (rs6000_legitimate_address_p): Ditto.
18372 (abi_v4_pass_in_fpr): Ditto.
18373 (setup_incoming_varargs): Ditto.
18374 (rs6000_gimplify_va_arg): Ditto.
18375 (rs6000_split_multireg_move): Ditto.
18376 (rs6000_savres_strategy): Ditto.
18377 (rs6000_emit_prologue_components): Ditto.
18378 (rs6000_emit_epilogue_components): Ditto.
18379 (rs6000_emit_prologue): Ditto.
18380 (rs6000_emit_epilogue): Ditto.
18381 (rs6000_elf_file_end): Ditto.
18382 (rs6000_function_value): Ditto.
18383 (rs6000_libcall_value): Ditto.
18384 * config/rs6000/rs6000.h: Ditto.
18385 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
18386 (TARGET_MINMAX): ... this. New.
18387 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
18388 * config/rs6000/rs6000.md: Remove Xilinx FP support.
18389 (*movsi_internal1_single): Delete.
18390 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
18391 mfpu=, mxilinx-fpu): Delete.
18392 * config/rs6000/singlefp.h: Delete.
18393 * config/rs6000/sysv4.h: Remove Xilinx FP support.
18394 * config/rs6000/t-rs6000: Ditto.
18395 * config/rs6000/t-xilinx: Delete.
18396 * config/rs6000/titan.md: Adjust for fp_type removal.
18397 * config/rs6000/vsx.md: Remove Xilinx FP support.
18398 (VStype_simple): Delete.
18399 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
18400 * config/rs6000/xfpu.h: Delete.
18401 * config/rs6000/xfpu.md: Delete.
18402 * config/rs6000/xilinx.h: Delete.
18403 * config/rs6000/xilinx.opt: Delete.
18404 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
18405 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
18407 2018-05-04 Tom de Vries <tom@codesourcery.com>
18410 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
18413 2018-05-04 Richard Biener <rguenther@suse.de>
18415 PR middle-end/85627
18416 * tree-complex.c (update_complex_assignment): We are always in SSA form.
18417 (expand_complex_div_wide): Likewise.
18418 (expand_complex_operations_1): Likewise.
18419 (expand_complex_libcall): Preserve EH info of the original stmt.
18420 (tree_lower_complex): Handle removed blocks.
18421 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
18422 on complex multiplication and division libcall builtins.
18424 2018-05-04 Richard Biener <rguenther@suse.de>
18426 PR middle-end/85574
18427 * fold-const.c (negate_expr_p): Restrict negation of operand
18428 zero of a division to when we know that can happen without
18430 (fold_negate_expr_1): Likewise.
18432 2018-05-04 Jakub Jelinek <jakub@redhat.com>
18435 * real.h (real_nextafter): Declare.
18436 * real.c (real_nextafter): New function.
18437 * fold-const-call.c (fold_const_nextafter): New function.
18438 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
18439 CASE_CFN_NEXTTOWARD.
18440 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
18441 even when arg1_mode is different from arg0_mode.
18443 2018-05-03 Nathan Sidwell <nathan@acm.org>
18445 * doc/extend.texi (Deprecated Features): Remove
18446 -ffriend-injection.
18447 (Backwards Compatibility): Likewise.
18448 * doc/invoke.texi (C++ Language Options): Likewise.
18449 (C++ Dialect Options): Likewise.
18451 2018-05-03 Jakub Jelinek <jakub@redhat.com>
18454 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
18455 _mm512_mask_mullox_epi64): New intrinsics.
18457 2018-05-03 Tom de Vries <tom@codesourcery.com>
18460 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
18461 dump files): Add offload-tree.
18463 2018-05-03 Richard Biener <rguenther@suse.de>
18465 PR tree-optimization/85615
18466 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
18467 to loops not nested in BBs loop father to avoid creating multi-entry
18470 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18472 PR tree-optimization/70291
18473 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
18474 arguments. Change return type to tree. Emit libcall as a new
18475 statement rather than replacing existing one when inplace_p is true.
18476 (expand_complex_multiplication_components): New function.
18477 (expand_complex_multiplication): Expand floating-point complex
18478 multiplication using the above.
18479 (expand_complex_division): Rename inner_type parameter to type.
18480 Update expand_complex_libcall call-site.
18481 (expand_complex_operations_1): Update expand_complex_multiplication
18482 and expand_complex_division call-sites.
18484 2018-05-02 Jakub Jelinek <jakub@redhat.com>
18487 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
18488 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
18489 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
18490 the highest significant bit of the shift count mask is clear. In
18491 check whether and[sq]i3 is needed verify that all significant bits
18492 of the shift count other than the highest are set.
18494 2018-05-02 Tom de Vries <tom@codesourcery.com>
18497 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
18498 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
18499 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
18500 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
18501 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
18502 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
18503 __builtin_goacc_parlevel_size.
18505 2018-05-02 Richard Biener <rguenther@suse.de>
18507 PR tree-optimization/85597
18508 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
18509 do not use split vect_get_vec_defs call but call vect_get_slp_defs
18512 2018-05-02 Tom de Vries <tom@codesourcery.com>
18515 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
18516 dump files): Add ltrans-tree.
18518 2018-05-02 Tom de Vries <tom@codesourcery.com>
18521 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
18522 dump files): Add wpa-ipa.
18524 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
18526 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
18527 powerpc*-*-linux*paired* target.
18528 * config/rs6000/750cl.h: Delete.
18529 * config/rs6000/paired.h: Delete.
18530 * config/rs6000/paired.md: Delete.
18531 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
18533 * config/rs6000/rs6000-builtin.def: Remove paired float support.
18534 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
18535 comment. Remove paired float support.
18536 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
18537 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
18539 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
18540 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
18542 * config/rs6000/rs6000.c: Remove paired float support.
18543 (paired_expand_vector_init, paired_expand_vector_move,
18544 paired_emit_vector_compare, paired_emit_vector_cond_expr,
18545 (paired_expand_lv_builtin, paired_expand_stv_builtin,
18546 paired_expand_builtin, paired_expand_predicate_builtin,
18547 paired_init_builtins): Delete.
18548 * config/rs6000/rs6000.h: Remove paired float support.
18549 * config/rs6000/rs6000.md: Remove paired float support.
18550 (move_from_CR_ov_bit): Delete.
18551 * config/rs6000/rs6000.opt (mpaired): Delete.
18552 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
18553 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
18555 2018-05-02 Richard Biener <rguenther@suse.de>
18557 PR middle-end/85567
18558 * gimplify.c (gimplify_save_expr): When in SSA form allow
18559 SAVE_EXPRs to compute to SSA vars.
18561 2018-05-02 Jakub Jelinek <jakub@redhat.com>
18564 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
18565 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
18566 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
18567 clobber operands[2], instead use a new pseudo. Formatting fixes.
18569 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
18571 PR tree-optimization/85586
18572 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
18573 exit early for statements in the same group if the accesses are
18576 2018-05-02 Tom de Vries <tom@codesourcery.com>
18579 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
18582 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
18584 PR tree-optimization/85143
18585 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
18587 2018-05-01 Tom de Vries <tom@codesourcery.com>
18590 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
18591 not found" error message.
18593 2018-05-01 Tom de Vries <tom@codesourcery.com>
18596 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
18597 * vec.c (test_ordered_remove_if): New function.
18598 (vec_c_tests): Call test_ordered_remove_if.
18599 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
18600 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
18601 * tree-vect-patterns.c (vect_pattern_recog_1): Use
18602 VEC_ORDERED_REMOVE_IF.
18604 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18606 PR tree-optimization/82665
18607 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
18608 pointer subtraction where arguments come from a memchr call.
18610 2018-05-01 Jakub Jelinek <jakub@redhat.com>
18612 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
18613 --push-state --as-needed and --pop-state instead of --as-needed and
18614 --no-as-needed if ld supports it.
18615 * configure: Regenerated.
18618 * doc/install.texi2html: Replace _002d with - and _002a with * in
18619 generated html files using sed.
18621 2018-04-30 David Malcolm <dmalcolm@redhat.com>
18624 * gcc-rich-location.c (blank_line_before_p): New function.
18625 (use_new_line): New function.
18626 (gcc_rich_location::add_fixit_insert_formatted): New function.
18627 * gcc-rich-location.h
18628 (gcc_rich_location::add_fixit_insert_formatted): New function.
18630 2018-04-30 David Malcolm <dmalcolm@redhat.com>
18632 * selftest.c (assert_streq): Rename "expected" and "actual" to
18633 "val1" and "val2". Extend NULL-handling to cover both inputs
18634 symmetrically, while still requiring both to be non-NULL for a pass.
18635 * selftest.h (assert_streq): Rename "expected" and "actual" to
18637 (ASSERT_EQ): Likewise.
18638 (ASSERT_EQ_AT): Likewise.
18639 (ASSERT_KNOWN_EQ): Likewise.
18640 (ASSERT_KNOWN_EQ_AT): Likewise.
18641 (ASSERT_NE): Likewise.
18642 (ASSERT_MAYBE_NE): Likewise.
18643 (ASSERT_MAYBE_NE_AT): Likewise.
18644 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
18645 the assertion to pass.
18646 (ASSERT_STREQ_AT): Likewise.
18648 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
18650 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
18651 interaction with -pie.
18653 2018-04-30 David Malcolm <dmalcolm@redhat.com>
18655 * selftest.h: Fix alphabetization of per-source-file selftest
18658 2018-04-30 Jason Merrill <jason@redhat.com>
18660 PR c++/61982 - dead stores to destroyed objects.
18661 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
18664 2018-04-30 Jason Merrill <jason@redhat.com>
18666 * tree.c (build_clobber): New.
18667 * tree.h: Declare it.
18668 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
18670 2018-04-30 David Malcolm <dmalcolm@redhat.com>
18672 * diagnostic-show-locus.c (layout::layout): Update for
18673 location_get_source_line returning a char_span.
18674 (struct char_span): Move to input.h.
18675 (struct correction): Update for fields in char_span becoming
18677 (struct source_line): Update for location_get_source_line
18678 returning a char_span.
18679 (layout::print_line): Likewise.
18680 * edit-context.c (edited_file::print_content): Likewise.
18681 (edited_file::print_diff_hunk): Likewise.
18682 (edited_file::print_run_of_changed_lines): Likewise.
18683 (edited_file::get_num_lines): Likewise.
18684 (edited_line::edited_line): Likewise.
18685 * final.c (asm_show_source): Likewise.
18686 * input.c (location_get_source_line): Convert return type
18687 from const char * to char_span, losing the final "line_len"
18689 (dump_location_info): Update for the above.
18690 (get_substring_ranges_for_loc): Likewise. Use a char_span
18691 when handling the literal within the line.
18692 (test_reading_source_line): Update for location_get_source_line
18693 returning a char_span.
18694 * input.h (class char_span): Move here from
18695 diagnostic-show-locus.c, converting from a struct to a class.
18696 Make data members private.
18697 (char_span::operator bool): New.
18698 (char_span::length): New.
18699 (char_span::get_buffer): New.
18700 (char_span::operator[]): New.
18701 (char_span::subspan): Make const.
18702 (char_span::xstrdup): New.
18703 (location_get_source_line): Convert return type from const char *
18704 to char_span, losing the final "line_size" param.
18706 2018-04-30 Jan Hubicka <jh@suse.cz>
18708 * lto-wrapper.c (ltrans_priorities): New static var.
18709 (cmp_priority): New.
18710 (run_gcc): Read priorities and if doing parallel build order
18711 the Makefile by them.
18713 2018-04-30 David Malcolm <dmalcolm@redhat.com>
18715 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
18717 2018-04-30 Richard Biener <rguenther@suse.de>
18719 * tree-cfg.c (verify_address): Remove base argument, add
18720 flag whether to check TREE_ADDRESSABLE and do that.
18721 (verify_expr): Remove.
18722 (verify_types_in_gimple_reference): Add pieces from verify_expr.
18723 (verify_gimple_assign_single): Likewise.
18724 (verify_gimple_switch): Likewise.
18725 (verify_expr_location_1): Dereference tp once. Add (disabled)
18726 piece from verify_expr.
18727 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
18729 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
18731 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
18733 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
18735 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
18736 (small_data_pattern): Likewise.
18737 (arc_rewrite_small_data): Likewise.
18738 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
18739 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
18740 (get_symbol_alignment): New function.
18741 (legitimate_small_data_address_p): Likewise.
18742 (legitimate_scaled_address): Update, call
18743 legitimate_small_data_address_p.
18744 (output_sdata): New static variable.
18745 (arc_print_operand): Update how we handle small data operands.
18746 (arc_print_operand_address): Likewise.
18747 (arc_legitimate_address_p): Update, use
18748 legitimate_small_data_address_p.
18749 (arc_rewrite_small_data_p): Remove.
18750 (arc_rewrite_small_data_1): Likewise.
18751 (arc_rewrite_small_data): Likewise.
18752 (small_data_pattern): Likewise.
18753 (compact_sda_memory_operand): Update to use
18754 legitimate_small_data_address_p and get_symbol_alignment.
18755 (prepare_move_operands): Don't rewite sdata pattern.
18756 (prepare_extend_operands): Remove.
18757 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
18759 (zero_extendqisi2): Likewise.
18760 (zero_extendhisi2): Likewise.
18761 (extendqihi2): Likewise.
18762 (extendqisi2): Likewise.
18763 (extendhisi2): Likewise.
18764 (addsi3): Likewise.
18765 (subsi3): Likewise.
18766 (andsi3): Likewise.
18767 * config/arc/constraints.md (Usd): Change it to memory constraint.
18769 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
18771 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
18772 as source of std instructions.
18773 * config/arc/arc.md (movsi_insn): Update pattern predicate to
18774 allow 6-bit constants as source for store instructions.
18775 (movdi_insn): Update instruction pattern to allow 6-bit constants
18776 as source for store instructions.
18778 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
18780 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
18782 2018-04-30 Nathan Sidwell <nathan@acm.org>
18783 Sandra Loosemore <sandra@codesourcery.com>
18785 * dumpfile.c (dump_open): Allow '-' for stdout.
18786 * doc/invoke.texi (Developer Options): Document dump filename
18787 determination early. Document stdin/stdout selection.
18789 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
18791 Microblaze Target: PIC data text relative
18793 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
18794 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
18796 * config/microblaze/microblaze.h (microblaze_constant_address_p):
18797 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
18798 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
18799 New addressing mode for data-text relative position indepenedent code.
18800 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
18801 'ADDRESS_SYMBOLIC_TXT_REL'.
18802 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
18803 (microblaze_legitimate_pic_operand): Exclude function calls from
18804 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
18805 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
18807 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
18808 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
18809 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
18810 for 'address + offset'.
18811 (microblaze_expand_prologue): Add new function prologue call for
18813 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
18814 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
18815 table in case of TARGET_PIC_DATA_TEXT_REL.
18816 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
18817 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
18818 Add new macros 'UNSPEC_TEXT',
18819 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
18820 + exclude function calls from 'UNSPEC_PLT' in case of data text
18822 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
18823 new target hook for generating address diff vector tables in case of
18825 * doc/tm.texi : Regenerate.
18826 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
18827 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
18828 of addr diff vector generation.
18829 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
18830 target hook definition.
18831 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
18832 Add default function for generate_pic_addr_diff_vec -> flag_pic.
18833 * doc/invoke.texi (Add new pic option): Add new microblaze pic
18834 option for data text relative.
18836 2018-04-30 Richard Biener <rguenther@suse.de>
18838 * tree-chrec.h (evolution_function_is_constant_p): Remove
18840 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
18842 2018-04-30 Richard Biener <rguenther@suse.de>
18845 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
18847 2018-04-30 Richard Biener <rguenther@suse.de>
18849 PR tree-optimization/28364
18850 PR tree-optimization/85275
18851 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
18852 copying first exit test.
18854 2018-04-28 Mark Wielaard <mark@klomp.org>
18856 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
18857 dwarf_version >= 5.
18858 (dwarf_AT): Handle DW_AT_addr_base.
18859 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
18861 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
18864 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
18865 (*ashl<dwi>3_doubleword_mask_1): Ditto.
18866 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
18867 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
18869 2018-04-28 Richard Biener <rguenther@suse.de>
18871 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
18872 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
18873 to reflect use. Only add interesting stmts.
18875 2018-04-27 Martin Jambor <mjambor@suse.cz>
18878 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
18879 the jump function allows for passing through aggregate values.
18881 2018-04-27 David Malcolm <dmalcolm@redhat.com>
18883 * input.h (in_system_header_at): Convert from macro to inline
18885 (from_macro_expansion_at): Likewise.
18886 (from_macro_definition_at): Likewise.
18888 2018-04-27 Jeff Law <law@redhat.com>
18890 * config.gcc: Mark tile* targets as deprecated/obsolete.
18892 2018-04-27 Richard Biener <rguenther@suse.de>
18894 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
18897 2018-04-27 Richard Biener <rguenther@suse.de>
18899 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
18901 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
18903 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
18904 with Yd constraint. Set "preferred_for_speed" attribute from
18905 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
18906 with Yd constraint.
18907 (*movdi_internal): Ditto.
18908 (movti_interunit splitters): Remove
18909 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
18910 (movdi_interunit splitters): Ditto.
18911 * config/i386/constraints.md (Ye): Remove.
18912 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
18914 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18917 * config/aarch64/constraints.md (Usg): Limit to 31.
18918 (Usj): Limit to 63.
18920 2018-04-27 Jakub Jelinek <jakub@redhat.com>
18922 PR tree-optimization/85529
18923 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
18924 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
18925 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
18926 zero extension or masking of the MSB bit.
18927 (optimize_range_tests): Add FIRST_BB argument, pass it through
18928 to optimize_range_tests_var_bound.
18929 (maybe_optimize_range_tests, reassociate_bb): Adjust
18930 optimize_range_tests callers.
18932 2018-04-26 Richard Biener <rguenther@suse.de>
18933 Jakub Jelinek <jakub@redhat.com>
18935 * cgraph.h (symbol_table): Just declare debug method here.
18936 * symtab.c (symbol_table::debug): Define.
18938 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
18940 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
18942 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
18944 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
18945 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
18946 (*movdi_internal): Substitute Yi and Yj constraint with x
18947 and Ym and Yn constraint with y constraint. Update "isa"
18948 attribute and set "preferred_for_speed" attribute from
18949 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
18950 (*movsi_internal): Ditto.
18951 (*movdf_internal): Ditto.
18952 (*movsf_internal): Ditto.
18953 (*zero_extendsidi2): Ditto.
18954 * config/i386/sse.md (vec_set<mode>_0): Ditto.
18955 (sse2_loadld): Ditto.
18956 (*vec_extract<ssevecmodelower>_0): Ditto.
18957 (*vec_extractv4si_0_zext_sse4): Ditto.
18958 (vec_concatv2di): Ditto.
18959 (*vec_dup<mode>): Ditto.
18960 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
18961 * config/i386/constraints.md (Yi): Remove.
18966 2018-04-26 Nathan Sidwell <nathan@acm.org>
18968 * dumpfile.c (dump_open): New.
18969 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
18970 (dump_finish): Detect stdio/stderr by value not name.
18972 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
18974 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
18976 2018-04-26 Tom de Vries <tom@codesourcery.com>
18979 * config/nvptx/nvptx.c (verify_neutering_jumps)
18980 (verify_neutering_labels): New function
18981 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
18983 2018-04-26 Tom de Vries <tom@codesourcery.com>
18986 * config/nvptx/nvptx.c (needs_neutering_p): New function.
18987 (nvptx_single): Use needs_neutering_p to skip over insns that do not
18990 2018-04-26 Richard Biener <rguenther@suse.de>
18991 Tom de Vries <tom@codesourcery.com>
18994 * lto-streamer-out.c (output_function): Fixup loops if required to match
18995 discovery done in the reader.
18997 2018-04-26 Richard Biener <rguenther@suse.de>
18999 PR tree-optimization/85116
19000 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
19001 have a loop exit from the single latch predecessor. Remove
19002 case of header with just condition.
19003 (ch_base::copy_headers): Exclude infinite loops from any
19005 (pass_ch::execute): Record exits.
19007 2018-04-26 Richard Biener <rguenther@suse.de>
19009 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
19010 prologue cost vector and pass it to vect_get_load_cost.
19011 (vect_get_peeling_costs_all_drs): Likewise.
19012 (vect_peeling_hash_get_lowest_cost): Likewise.
19013 (vect_enhance_data_refs_alignment): Likewise.
19015 2018-04-26 Richard Biener <rguenther@suse.de>
19017 PR middle-end/85450
19018 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
19019 checking of integer<->pointer conversions.
19020 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
19021 sign-/zero-extending pointer types.
19022 (expand_omp_for_static_chunk): Likewise.
19024 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
19025 Jean Lee <xiaoyur347@gmail.com>
19027 * config/mips/mips.c (mips_asan_shadow_offset): New function.
19028 (TARGET_ASAN_SHADOW_OFFSET): Define.
19029 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
19030 true for -fsanitize=address.
19032 2018-04-25 Mark Wielaard <mark@klomp.org>
19034 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
19037 2018-04-25 Jakub Jelinek <jakub@redhat.com>
19039 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
19040 than "alu", remove explicit "memory" and "imm_disp" attributes.
19041 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
19043 PR middle-end/85414
19044 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
19045 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
19046 gen_lowpart_no_emit.
19048 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
19051 * config/i386/i386.c (ix86_expand_builtin): Change memory
19052 operand to XI, extend p0 to Pmode.
19053 * config/i386/i386.md: Change unspec volatile and operand
19054 1 mode to XI, change operand 0 mode to P.
19056 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
19058 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
19059 GET_MODE_MASK before any checking.
19060 (nds32_can_use_bset_p): Likewise.
19061 (nds32_can_use_btgl_p): Likewise.
19063 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
19065 * config/nds32/nds32-doubleword.md: New define_split pattern for
19066 illegal register number.
19068 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
19070 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
19072 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
19074 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
19076 2018-04-25 Richard Biener <rguenther@suse.de>
19078 * lto-streamer.h (LTO_major_version): Bump to 8.
19080 2018-04-25 Jakub Jelinek <jakub@redhat.com>
19082 * BASE-VER: Set to 9.0.0.
19084 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
19086 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
19087 in __abskf2 and __powikf2.
19089 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19092 * config/aarch64/constraints.md (Usg, Usj): New constraints.
19093 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
19094 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
19095 Use the above on operand 2. Reindent.
19096 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
19098 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
19101 * common/config/i386/i386-common.c (ix86_handle_option): Don't
19103 * config/i386/i386.opt (mcet): Removed.
19104 * doc/install.texi: Remove -mcet documentation.
19105 * doc/invoke.texi: Likewise.
19107 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
19110 * doc/install.texi: Remove -mcet from bootstrap-cet.
19112 2018-04-24 Jakub Jelinek <jakub@redhat.com>
19115 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
19116 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
19120 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
19121 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
19122 containing a CONST_VECTOR.
19124 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
19126 * doc/install.texi: Update newlib dependency for nvptx.
19128 2018-04-24 Jakub Jelinek <jakub@redhat.com>
19131 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
19132 instead of INTVAL when shifting x left.
19134 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
19136 PR tree-optimization/85478
19137 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
19138 vect_grouped_store_supported for single element vectors.
19140 2018-04-24 Richard Biener <rguenther@suse.de>
19143 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
19144 load cost increase to the case of non-constant step.
19146 2018-04-24 Jakub Jelinek <jakub@redhat.com>
19149 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
19150 destination if any_malformed_asm.
19152 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
19154 PR middle-end/85496
19155 * expr.c (store_field): In the bitfield case, if the value comes from
19156 a function call and is returned in registers by means of a PARALLEL,
19157 do not change the mode of the temporary unless BLKmode and VOIDmode.
19159 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
19161 PR rtl-optimization/85423
19162 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
19163 dependencies to debug insns when the previous insn is non-debug.
19165 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
19167 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
19168 enums into a single definition.
19169 (fls): Fix predicates and printing.
19172 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
19174 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
19175 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
19176 and short u6 immediate.
19177 (check_if_valid_sleep_operand): Remove.
19178 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
19180 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
19182 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
19183 flag_always_save_lp condition.
19184 * config/nds32/nds32.opt (malways-save-lp): New option.
19186 2018-04-22 Shiva Chen <shiva0217@gmail.com>
19188 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
19189 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
19190 * config/nds32/nds32.h
19191 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
19192 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
19194 2018-04-22 Shiva Chen <shiva0217@gmail.com>
19196 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
19197 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
19199 2018-04-22 Shiva Chen <shiva0217@gmail.com>
19200 Chung-Ju Wu <jasonwucj@gmail.com>
19202 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
19204 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
19205 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
19207 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
19209 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
19211 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
19213 * config/nds32/nds32-protos.h (nds32_data_alignment,
19214 nds32_local_alignment): Declare.
19215 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
19216 nds32_local_alignment): New functions.
19217 (TARGET_CONSTANT_ALIGNMENT): Define.
19218 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
19220 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
19222 * config/nds32/nds32.c
19223 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
19224 (TARGET_MODES_TIEABLE_P): Likewise.
19226 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
19228 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
19229 level Ofast and Og.
19231 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
19232 Chung-Ju Wu <jasonwucj@gmail.com>
19234 * config/nds32/constants.md (unspec_volatile_element): Add enum values
19235 for unaligned access.
19236 * config/nds32/nds32-intrinsic.c: Implementation of expanding
19238 * config/nds32/nds32-intrinsic.md: Likewise.
19239 * config/nds32/nds32_intrinsic.h: Likewise.
19240 * config/nds32/nds32.h (nds32_builtins): Likewise.
19241 * config/nds32/nds32.opt (munaligned-access): New option.
19242 * config/nds32/nds32.c (nds32_asm_file_start): Display
19243 flag_unaligned_access status.
19245 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
19247 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
19248 -mno-relax is present.
19249 * config/riscv/linux.h (LINK_SPEC): Ditto.
19251 2018-04-20 Martin Sebor <msebor@redhat.com>
19254 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
19256 (gimple_fold_builtin_stxcpy_chk): Same.
19257 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
19259 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
19262 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
19263 __powikf2 when long double is IEEE 128-bit.
19265 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
19267 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
19268 step to make sure stack always aligned.
19270 2018-04-20 Carl Love <cel@us.ibm.com>
19273 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
19274 size check for arg0.
19276 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
19277 Tom de Vries <tom@codesourcery.com>
19280 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
19281 Emit insns for calls too.
19282 (nvptx_find_par): Always look for worker-level predecessor insn.
19283 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
19285 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
19286 (nvptx_process_pars): Propagate frames for calls.
19288 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
19291 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
19293 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
19294 (ix86_handle_option): Don't handle OPT_mibt.
19295 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
19297 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
19299 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
19300 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
19301 (ix86_target_macros): Define __CET__ with flag_cf_protection
19302 for -fcf-protection.
19303 * config/i386/i386.c (isa2_opts): Remove -mibt.
19304 * config/i386/i386.h (TARGET_IBT): Removed.
19305 (TARGET_IBT_P): Likewise.
19306 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
19307 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
19308 * config/i386/i386.opt (mcet): Update help message.
19309 (mshstk): Likewise.
19311 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
19312 -mcet as an alias for -mshstk.
19314 2018-04-20 Richard Biener <rguenther@suse.de>
19316 PR middle-end/85475
19317 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
19318 complexity by forcing a single use of the multiply operand.
19320 2018-04-20 Martin Jambor <mjambor@suse.cz>
19323 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
19324 recursion dependency to only apply to non-clones.
19326 2018-04-20 Martin Jambor <mjambor@suse.cz>
19329 * ipa-cp.c (create_specialized_node): Check that clones of
19330 self-recursive edges exist during IPA-CP.
19332 2018-04-19 Toon Moene <toon@moene.org>
19334 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
19337 2018-04-19 Jakub Jelinek <jakub@redhat.com>
19339 PR tree-optimization/85467
19340 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
19341 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
19342 VECTOR_CST element to type.
19344 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
19347 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
19348 * config/i386/i386.md (builtin_setjmp_setup): Removed.
19349 (builtin_longjmp): Likewise.
19350 (save_stack_nonlocal): New pattern.
19351 (restore_stack_nonlocal): Likewise.
19353 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
19356 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
19357 Replace ASM_OUTPUT_LABEL with fprintf.
19359 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
19362 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
19363 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
19364 * config/i386/i386-c.c (ix86_target_macros_internal): Also
19365 define __IBT__ and __SHSTK__ for -fcf-protection.
19366 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
19368 (ix86_trampoline_init): Likewise.
19369 (x86_output_mi_thunk): Likewise.
19370 (ix86_notrack_prefixed_insn_p): Likewise.
19371 (ix86_option_override_internal): Don't disallow -fcf-protection.
19372 * config/i386/i386.md (rdssp<mode>): Also enable for
19374 (incssp<mode>): Likewise.
19375 (nop_endbr): Likewise.
19376 * config/i386/i386.opt (mcet): Change help message to built-in
19379 (mshstk): Likewise.
19380 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
19381 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
19382 enable CET built-in functions.
19384 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
19386 * common/config/i386/i386-common.c
19387 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
19388 OPTION_MASK_ISA_MOVDIRI_UNSET,
19389 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
19390 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
19391 * config.gcc (movdirintrin.h): New header.
19392 * config/i386/cpuid.h (bit_MOVDIRI,
19393 bit_MOVDIR64B): New bits.
19394 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
19396 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
19397 (VOID, PVOID, PCVOID)): New function types.
19398 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
19399 __builtin_ia32_directstoreu_u64,
19400 __builtin_ia32_movdir64b): New builtins.
19401 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
19402 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
19404 (ix86_valid_target_attribute_inner_p): Ditto.
19405 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
19406 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
19407 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
19408 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
19409 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
19410 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
19411 (movdiri<mode>, movdir64b_<mode>): New.
19412 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
19413 * config/i386/immintrin.h: Include movdirintrin.h.
19414 * config/i386/movdirintrin.h: New file.
19415 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
19417 2018-04-19 Richard Biener <rguenther@suse.de>
19419 PR middle-end/85455
19420 * cfg.c (clear_bb_flags): When loop state says we have
19421 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
19423 2018-04-19 Richard Biener <rguenther@suse.de>
19425 PR tree-optimization/84737
19426 * tree-vect-data-refs.c (vect_copy_ref_info): New function
19427 copying restrict info.
19428 (vect_setup_realignment): Use it.
19429 * tree-vectorizer.h (vect_copy_ref_info): Declare.
19430 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
19431 the first DR to all generated stores.
19432 (vectorizable_load): Likewise for loads.
19434 2018-04-19 Jakub Jelinek <jakub@redhat.com>
19436 PR tree-optimization/85446
19437 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
19438 the integral and pointer types to have the same precision.
19440 * doc/install.texi: Document --disable-cet being the default and
19443 2018-04-18 Martin Liska <mliska@suse.cz>
19445 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
19448 2018-04-18 Martin Liska <mliska@suse.cz>
19451 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
19454 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
19455 arguments if they are comparable.
19457 2018-04-18 Martin Liska <mliska@suse.cz>
19460 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
19463 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
19466 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
19469 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
19470 ENDBR after calling __morestack.
19472 2018-04-18 David Malcolm <dmalcolm@redhat.com>
19475 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
19476 by using gcc_base_ver to generate a gcc_driver_version, and use
19477 it when generating GCC_DRIVER_NAME.
19478 * configure: Regenerate.
19480 2018-04-18 Jakub Jelinek <jakub@redhat.com>
19483 * config.gcc: Obsolete powerpc*-*-*spe*.
19485 2018-04-17 Jakub Jelinek <jakub@redhat.com>
19488 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
19489 (stabstr_D): Change type of unum from unsigned int to
19490 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
19493 2018-04-17 Jim Wilson <jimw@sifive.com>
19496 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
19497 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
19498 Set arg_pointer_offset after using pretend_args_size.
19500 2018-04-17 Jakub Jelinek <jakub@redhat.com>
19502 PR rtl-optimization/85431
19503 * dse.c (record_store): Ignore zero width stores.
19506 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
19507 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
19508 __builtin_stack_restore rather than after it.
19509 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
19510 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
19511 argument instead of virtual_dynamic_stack_rtx.
19513 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
19515 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
19517 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19518 Add note to error message to explain internal mapping of overloaded
19519 built-in function name to non-overloaded built-in function name.
19520 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
19523 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
19526 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
19527 where the inputs overlap with the output.
19529 2018-04-17 Jakub Jelinek <jakub@redhat.com>
19531 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
19532 (=v, v) alternative and explicit "memory" attribute.
19533 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
19534 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
19536 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
19537 "sselog1" type instead of "sselog".
19538 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
19539 "sselog". Remove explicit "memory" attribute.
19540 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
19541 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
19543 (vec_extract_hi_v32hi): Merge all alternatives into one, use
19544 "sselog1" type instead of "sselog". Remove explicit "memory"
19546 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
19547 use "sselog1" type instead of "sselog". Remove explicit "memory"
19549 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
19550 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
19552 (vec_extract_hi_v64qi): Merge all alternatives into one, use
19553 "sselog1" type instead of "sselog". Remove explicit "memory"
19555 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
19556 use "sselog1" type instead of "sselog". Remove explicit "memory"
19560 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
19562 PR middle-end/85414
19563 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
19566 2018-04-17 Martin Jambor <mjambor@suse.cz>
19569 * ipa-cp.c (create_specialized_node): Call
19570 expand_all_artificial_thunks if necessary.
19572 2018-04-17 Martin Liska <mliska@suse.cz>
19575 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
19576 in message, remote space in between '_G' and '('.
19578 2018-04-17 Jakub Jelinek <jakub@redhat.com>
19581 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
19582 avx512f_vmcmp<mode>3<round_saeonly_name>,
19583 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
19584 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
19585 avx512f_rndscale<mode><round_saeonly_name>,
19586 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
19587 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
19588 Use %<iptr>2 instead of %2 for -masm=intel.
19589 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
19590 avx512f_vcvttss2usi<round_saeonly_name>,
19591 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
19593 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
19594 avx512f_vcvttsd2usi<round_saeonly_name>,
19595 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
19596 Use %q1 instead of %1 for -masm=intel.
19597 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
19598 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
19599 of %3 for -masm=intel.
19600 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
19602 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
19604 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
19606 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
19607 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
19609 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
19611 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
19612 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
19613 %g1 and one with %0 and %1.
19614 (avx512er_vmrcp28<mode><round_saeonly_name>,
19615 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
19616 %1 for -masm=intel.
19617 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
19618 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
19619 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
19620 of %0 and %{%4%} for -masm=intel.
19621 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
19622 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
19623 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
19624 order of %0 and %{%5%}%{z%} for -masm=intel.
19626 2018-04-17 Jan Hubicka <jh@suse.cz>
19629 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
19631 2018-04-17 Martin Liska <mliska@suse.cz>
19634 * multiple_target.c (create_dispatcher_calls): Set apostrophes
19635 for target_clone error message. Make default implementation
19636 clone to be a local declaration.
19637 (separate_attrs): Add new argument and check for an empty
19639 (expand_target_clones): Handle it.
19640 (ipa_target_clone): Make redirection just for target_clones
19643 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
19644 Tom de Vries <tom@codesourcery.com>
19646 PR middle-end/84955
19647 * omp-expand.c (expand_oacc_for): Add dummy false branch for
19648 tiled basic blocks without omp continue statements.
19650 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
19653 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
19654 vec_extract expression as having side effects to make sure it gets
19657 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
19660 * config/i386/i386.c (get_builtin_code_for_version): Check
19663 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
19666 * config.gcc: Support "skylake".
19667 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
19669 * config/i386/i386.c (m_SKYLAKE): Define.
19670 (processor_target_table): Add "skylake".
19671 (ix86_option_override_internal): Add "skylake".
19672 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
19673 PROCESSOR_CANNONLAKE.
19674 (get_builtin_code_for_version): Fix priority for
19675 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
19676 PROCESSOR_SKYLAKE-AVX512.
19677 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
19678 (processor_type): Add PROCESSOR_SKYLAKE.
19680 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
19681 Jason Merrill <jason@redhat.com>
19684 * convert.c (convert_to_integer_1): Use direct recursion for
19685 enumeral types and types with a precision less than the number
19686 of bits in their mode.
19688 2018-04-16 Julia Koval <julia.koval@intel.com>
19691 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
19692 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
19694 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
19697 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
19698 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
19699 and -mno-direct-move.
19701 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
19704 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
19705 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
19706 Ensure negative shifts result in {0}.
19708 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
19710 PR rtl-optimization/79916
19711 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
19712 regs (if any) to define how to gnerate SD moves when LRA is in
19715 2018-04-13 Jakub Jelinek <jakub@redhat.com>
19717 PR rtl-optimization/85393
19718 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
19719 * except.c (expand_dw2_landing_pad_for_region): Make static.
19720 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
19721 a label and unconditional jump to old_bb, rather than
19722 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
19725 PR rtl-optimization/85376
19726 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
19727 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
19728 instead of a specific value.
19730 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
19731 Bin Cheng <bin.cheng@arm.com>
19733 PR tree-optimization/82965
19734 PR tree-optimization/83991
19735 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
19736 by_profile_only parameter.
19737 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
19738 information if the loop was predicted to iterate too many times.
19739 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
19741 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
19744 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
19747 2018-04-13 Martin Liska <mliska@suse.cz>
19748 Jakub Jelinek <jakub@redhat.com>
19750 PR middle-end/81657
19751 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
19752 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
19753 * builtins.c (expand_builtin_memory_copy_args): Use
19754 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
19755 handle dest_addr == pc_rtx.
19757 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
19760 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
19761 asked to not generate direct moves.
19762 (fix_trunc<mode>si2_stfiwx): Similar.
19763 (fix_trunc<mode>si2_internal): Similar.
19765 2018-04-12 Jakub Jelinek <jakub@redhat.com>
19768 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
19769 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
19770 lookup if dest in some wider mode is known to be const0_rtx and
19771 if so, record permanent equivalence for it to be ZERO_EXTEND of
19772 the narrower mode destination.
19774 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
19776 * lto-streamer-out.c (output_function): Revert 259346.
19777 * omp-expand.c (expand_oacc_for): Likewise.
19779 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
19781 PR rtl-optimization/85354
19782 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
19783 * sel-sched.c (sel_global_init): ... here.
19785 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
19788 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
19789 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
19790 mode for PE-COFF targets.
19791 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
19792 (i386_pe_asm_lto_end): Likewise.
19793 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
19794 (TARGET_ASM_LTO_END): Likewise.
19795 * config/i386/winnt.c (saved_debug_info_level): New static variable.
19796 (i386_pe_asm_lto_start): New function.
19797 (i386_pe_asm_lto_end): Likewise.
19799 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
19800 Richard Biener <rguenther@suse.de>
19802 PR middle-end/84955
19803 * lto-streamer-out.c (output_function): Fix CFG loop state before
19805 * omp-expand.c (expand_oacc_for): Handle calls to internal
19806 functions like regular functions.
19808 2018-04-12 Richard Biener <rguenther@suse.de>
19811 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
19812 for the early LTO debug to properly generate references to it
19813 during DIE emission. Do not re-use that for the skeleton for
19815 (dwarf2out_early_finish): Likewise.
19817 2018-04-12 Jakub Jelinek <jakub@redhat.com>
19820 * config/i386/sse.md
19821 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
19822 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
19823 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
19824 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
19825 and output is a reg, avoid creating invalid lowpart subreg, but
19826 instead split into a 512-bit move. Don't split if not AVX512VL,
19827 input is xmm16+ reg and output is a mem.
19828 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
19829 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
19830 xmm16+ reg and output is a mem.
19832 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19834 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
19835 also for flag_dwarf2_cfi_asm.
19837 2018-04-12 Jakub Jelinek <jakub@redhat.com>
19839 PR rtl-optimization/85342
19840 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
19841 a bool scalar var inside of the loop instead. Don't try to update
19842 recog_data.operand after failed apply_change_group.
19844 2018-04-12 Tom de Vries <tom@codesourcery.com>
19847 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
19848 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
19849 array with flexible array member as array without given dimension.
19850 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
19851 argument for undefined param to true.
19853 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
19856 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
19857 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
19858 from PowerPC section.
19859 * config/rs6000/sysv4.opt (mcall-): Improve help text.
19860 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
19861 help text that is too long.
19862 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
19863 help text that is too long.
19864 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
19865 help text that is too long.
19867 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
19869 * config/alpha/alpha.md (stack_probe_internal): Rename
19870 from "probe_stack". Update all callers.
19872 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
19874 PR rtl-optimization/84566
19875 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
19876 sched_macro_fuse_insns.
19878 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
19881 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
19882 (compute_block_dependences): ... from here.
19884 2018-04-11 Jakub Jelinek <jakub@redhat.com>
19886 PR tree-optimization/85331
19887 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
19888 from int to HOST_WIDE_INT.
19890 2018-04-11 Martin Jambor <mjambor@suse.cz>
19893 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
19894 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
19895 not the same as the source val.
19896 (cgraph_edge_brings_value_p): New parameter.
19897 (gather_edges_for_value): Pass destination value to
19898 cgraph_edge_brings_value_p.
19899 (perhaps_add_new_callers): Likewise.
19900 (get_info_about_necessary_edges): Likewise and exclude values brought
19901 only by self-recursive edges.
19902 (create_specialized_node): Redirect only clones of self-calling edges.
19903 (+self_recursive_pass_through_p): New function.
19904 (find_more_scalar_values_for_callers_subset): Use it.
19905 (find_aggregate_values_for_callers_subset): Likewise.
19906 (known_aggs_to_agg_replacement_list): Removed.
19907 (decide_whether_version_node): Re-calculate known constants for all
19908 remaining context clones.
19910 2018-04-11 Richard Biener <rguenther@suse.de>
19913 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
19914 from early DWARF output.
19915 (dwarf2out_early_finish): Output line info unconditionally into
19916 early DWARF and add reference to it.
19918 2018-04-11 Jakub Jelinek <jakub@redhat.com>
19921 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
19922 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
19923 other than V2DFmode using iptr mode attribute.
19924 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
19926 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
19928 PR rtl-optimization/84659
19929 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
19931 2018-04-11 Jakub Jelinek <jakub@redhat.com>
19934 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
19936 (output_loc_list): Pass address of a dummy size variable even in the
19937 locview handling loop.
19938 (index_location_lists): Add comment on why skip_loc_list_entry can't
19941 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
19944 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
19947 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
19950 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
19951 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
19952 and -mstring-compare-inline-limit.
19954 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
19957 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
19958 for stack clash protection in a register whenever we need it to be in
19961 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
19963 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
19964 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
19966 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
19969 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
19971 (mlong-double-): Ditto.
19972 * config/rs6000/sysv4.opt (msdata=): Ditto.
19973 (mtls-size=): Ditto.
19975 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19977 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
19978 erroneous entries for
19979 "vector int vec_ldl (int, long int *)", and
19980 "vector unsigned int vec_ldl (int, unsigned long int *)".
19981 Add comments and entries for
19982 "vector bool char vec_ldl (int, bool char *)",
19983 "vector bool short vec_ldl (int, bool short *)",
19984 "vector bool int vec_ldl (int, bool int *)",
19985 "vector bool long long vec_ldl (int, bool long long *)",
19986 "vector pixel vec_ldl (int, pixel *)",
19987 "vector long long vec_ldl (int, long long *)",
19988 "vector unsigned long long vec_ldl (int, unsigned long long *)".
19989 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
19990 type tree bool_long_long_type_node and correct definition of
19991 bool_V2DI_type_node to make reference to this new type tree.
19992 (rs6000_mangle_type): Replace erroneous reference to
19993 bool_long_type_node with bool_long_long_type_node.
19994 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
19995 comments to emphasize sign distinctions for char and int types and
19996 replace RS6000_BTI_bool_long constant with
19997 RS6000_BTI_bool_long_long constant. Also add comment to restrict
19998 use of RS6000_BTI_pixel.
19999 (bool_long_type_node): Remove this macro definition.
20000 (bool_long_long_type_node): New macro definition
20002 2018-04-10 Jakub Jelinek <jakub@redhat.com>
20004 PR rtl-optimization/85300
20005 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
20006 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
20007 simplify_unary_operation fails.
20009 2018-04-10 Martin Liska <mliska@suse.cz>
20011 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
20012 cgraph_edge and ipa_ref.
20014 2018-04-10 Jakub Jelinek <jakub@redhat.com>
20018 * config/i386/sse.md
20019 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
20020 computation of the VEC_MERGE selector from mask.
20021 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
20022 Fix decoding of the VEC_MERGE selector into mask.
20024 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
20026 PR tree-optimization/85286
20027 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
20029 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
20031 * final.c (final_1): Set insn_last_address as well as
20032 insn_current_address.
20034 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20037 * explow.c (emit_stack_probe): Call validize_mem on memory location
20038 before passing it to gen_probe_stack. Create address operand and
20039 legitimize it for the probe_stack_address case.
20041 2018-04-09 Jan Hubicka <jh@suse.cz>
20044 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
20045 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
20046 * tree.c (free_lang_data_in_type): Fix handling of binfos;
20048 (free_lang_data): Rebuild type inheritance graph.
20050 2018-04-09 Martin Sebor <msebor@redhat.com>
20052 * invoke.texi (-finline-small-functions): Mention other optimization
20054 (-findirect-inlining, -fpartial-inlining): Same.
20055 (-finline-functions-called-once): Same.
20056 (-freorder-blocks-and-partition): Same.
20058 2018-04-09 Jan Hubicka <jh@suse.cz>
20061 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
20062 jumps; choose last target that matches the criteria (i.e.
20063 no partition changes for non-crossing jumps).
20064 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
20065 support for redirecting crossing jumps to non-crossing.
20067 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
20069 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
20070 also for naked functions.
20072 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
20074 * config/arc/arc.md (add_shift): New pattern.
20075 (add_shift2): Likewise.
20076 (sub_shift): Likewise.
20077 (sub_shift_cmp0_noout): Likewise.
20078 (compare_si_ashiftsi): Likewise.
20079 (xbfu_cmp0_noout): New combine pattern.
20080 (xbfu_cmp0"): Likewise.
20081 (movsi_set_cc_insn): Place the predicable variant first.
20082 (commutative_binary_cmp0_noout): Remove clobber.
20083 (commutative_binary_cmp0): New pattern.
20084 (noncommutative_binary_cmp0): Likewise.
20085 (noncommutative_binary_cmp0_noout): Likewise.
20086 (noncommutative_binary_comparison_result_used): Removed.
20087 (rsub_cmp0): New pattern.
20088 (rsub_cmp0_noout): Likewise.
20089 (extzvsi): Changed, keep only meaningful variants.
20090 (SQH, SEZ): New iterators.
20091 (SQH_postfix): New mode attribute.
20092 (SEZ_prefix): New code attribute.
20093 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
20094 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
20095 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
20096 of numerical value.
20097 (noncommutative_operator): Check the availability of barrel
20100 2018-04-09 Richard Biener <rguenther@suse.de>
20102 PR tree-optimization/85284
20103 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
20104 Only use the niter constraining form of simple_iv when the exit
20105 is always executed.
20107 2018-04-09 Tom de Vries <tom@codesourcery.com>
20110 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
20111 (define_expand "*memory_barrier"): New define_expand.
20112 (define_insn "memory_barrier"): New insn.
20114 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
20116 PR rtl-optimization/80463
20117 PR rtl-optimization/83972
20118 PR rtl-optimization/83480
20120 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
20121 correct producer for the insn.
20122 (tidy_control_flow): Fixup seqnos in case of debug insns.
20124 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
20126 PR rtl-optimization/83913
20128 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
20129 different sched-times when merging exprs.
20131 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
20133 PR rtl-optimization/83962
20135 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
20136 tidy_fallthru_edge and tidy_control_flow.
20138 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
20140 PR rtl-optimization/83530
20142 * sel-sched.c (force_next_insn): New global variable.
20143 (remove_insn_for_debug): When force_next_insn is true, also leave only
20144 next insn in the ready list.
20145 (sel_sched_region): When the region wasn't scheduled, make another pass
20146 over it with force_next_insn set to 1.
20148 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
20150 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
20152 * config/nds32/constants.md (unspec_volatile_element): Add enum values
20153 for interrupt control.
20154 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
20155 functions for interrupt control.
20156 * config/nds32/nds32-intrinsic.md: Likewise.
20157 * config/nds32/nds32_intrinsic.h: Likewise.
20158 * config/nds32/nds32.h (nds32_builtins): Likewise.
20160 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
20162 * config/nds32/nds32.c (nds32_init_machine_status,
20163 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
20164 strict_aligned_p field.
20165 (nds32_expand_to_rtl_hook): New function.
20166 (TARGET_EXPAND_TO_RTL_HOOK): Define.
20167 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
20169 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
20170 Chung-Ju Wu <jasonwucj@gmail.com>
20172 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
20173 * config/nds32/nds32-n7.md: New file.
20174 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
20175 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
20177 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
20178 * config/nds32/nds32.md (pipeline_model): Add n7.
20179 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
20180 * config/nds32/pipelines.md: Include n7 settings.
20182 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
20183 Chung-Ju Wu <jasonwucj@gmail.com>
20185 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
20186 * config/nds32/nds32-e8.md: New file.
20187 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
20188 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
20190 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
20191 * config/nds32/nds32.md (pipeline_model): Add e8.
20192 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
20193 * config/nds32/pipelines.md: Include e8 settings.
20195 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
20196 Chung-Ju Wu <jasonwucj@gmail.com>
20198 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
20199 * config/nds32/nds32-n8.md: New file.
20200 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
20201 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
20203 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
20204 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
20205 * config/nds32/nds32.md (pipeline_model): Add n8.
20206 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
20207 * config/nds32/pipelines.md: Include n8 settings.
20209 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
20210 Chung-Ju Wu <jasonwucj@gmail.com>
20212 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
20213 * config/nds32/nds32-n9-2r1w.md: New file.
20214 * config/nds32/nds32-n9-3r2w.md: New file.
20215 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
20216 nds32_register_ports): New or modify for cpu n9.
20217 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
20219 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
20220 * config/nds32/nds32-utils.c: New file.
20221 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
20222 TARGET_MUL_SLOW): Define.
20223 * config/nds32/nds32.md (pipeline_model): New attribute.
20224 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
20225 New options that support cpu n9.
20226 * config/nds32/pipelines.md: Include n9 settings.
20227 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
20229 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
20231 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
20232 information if necessary.
20233 (output_cond_branch_compare_zero): Likewise.
20234 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
20235 (nds32_target_alignment): Refine for alignment.
20236 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
20237 (FUNCTION_BOUNDARY): Modify.
20238 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
20240 * config/nds32/nds32.opt (malways-align, malign-functions): New.
20242 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
20244 * config/nds32/constants.md (unspec_volatile_element): Add values for
20245 TLB operation and data prefetch.
20246 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
20247 functions for TLB operation and data prefetch.
20248 * config/nds32/nds32-intrinsic.md: Likewise.
20249 * config/nds32/nds32_intrinsic.h: Likewise.
20250 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
20251 (nds32_print_operand): Likewise.
20252 * config/nds32/nds32.h (nds32_builtins): Likewise.
20254 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
20255 Andrew Pinski <pinsika@gcc.gnu.org>
20257 PR middle-end/82976
20258 * match.pd: Use constant_boolean_node of correct type instead of
20259 boolean_true_node or boolean_false_node for simplifying
20260 pointer comparisons to zero.
20262 2018-04-07 Jakub Jelinek <jakub@redhat.com>
20264 PR tree-optimization/80021
20265 * tree.c (verify_type_variant): Make error call in verify_variant_match
20266 translatable and remove final full stop.
20268 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
20270 * config/nds32/constants.md (unspec_volatile_element): Add
20271 UNSPEC_VOLATILE_EH_RETURN.
20272 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
20273 nds32_output_stack_pop): Support dwarf exception handling process.
20274 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
20275 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
20276 exception handling process.
20277 (nds32_compute_stack_frame): Likewise.
20278 (nds32_return_addr_rtx): Likewise.
20279 (nds32_initial_elimination_offset): Likewise.
20280 (nds32_expand_prologue): Likewise.
20281 (nds32_expand_epilogue): Likewise.
20282 (nds32_dynamic_chain_address): New function.
20283 * config/nds32/nds32.h (machine_function): Add fields for dwarf
20284 exception handling.
20285 (DYNAMIC_CHAIN_ADDRESS): Define.
20286 (EH_RETURN_DATA_REGNO): Define.
20287 (EH_RETURN_STACKADJ_RTX): Define.
20288 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
20289 patterns for dwarf exception handling.
20291 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
20293 * config/nds32/nds32.h: Clean up obsolete macros.
20295 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
20297 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
20298 Add enum values for particular instructions.
20299 * config/nds32/nds32-intrinsic.c: Implementation of expanding
20300 particular intrinsic functions.
20301 * config/nds32/nds32-intrinsic.md: Likewise.
20302 * config/nds32/nds32_intrinsic.h: Likewise.
20303 * config/nds32/nds32.h (nds32_builtins): Likewise.
20304 * config/nds32/nds32.md (type): Add pbsad and pbsada.
20305 (btst, ave): New patterns for particular instructions.
20307 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
20309 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
20310 Add enum values for atomic load/store and memory sync.
20311 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
20313 * config/nds32/nds32-intrinsic.md: Likewise.
20314 * config/nds32/nds32_intrinsic.h: Likewise.
20315 * config/nds32/nds32.h (nds32_builtins): Likewise.
20317 2018-04-07 Jakub Jelinek <jakub@redhat.com>
20319 PR tree-optimization/85257
20320 * fold-const.c (native_encode_vector): If not all elts could fit
20321 and off is -1, return 0 rather than offset.
20322 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
20323 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
20324 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
20325 adjust buffer in native_interpret_expr call.
20327 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
20329 * config/nds32/constants.md (unspec_volatile_element): Add cache
20330 control enum values.
20331 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
20332 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
20333 * config/nds32/nds32.c (nds32_cctl_names): New.
20334 (nds32_print_operand): Handle cache control register names.
20335 * config/nds32/nds32.h (nds32_builtins): New enum values.
20336 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
20338 * config/nds32/nds32.md (type): Add mmu.
20339 * config/nds32/pipelines.md (simple_insn): Add mmu.
20341 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
20343 * config/nds32/nds32.md (type): Remove call.
20344 * config/nds32/pipelines.md (simple_insn): Likewise.
20346 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
20348 * config/nds32/constants.md (unspec_volatile_element): Add
20349 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
20350 UNSPEC_VOLATILE_FMFCFG.
20351 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
20352 description for fmfcfg and fmfcsr.
20353 (bdesc_1arg): Add fmtcsr.
20354 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
20355 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
20356 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
20357 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
20358 unspec_fmfcfg): New patterns.
20359 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
20360 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
20361 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
20362 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
20363 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
20364 __nds32__fmfcfg): Define.
20366 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
20368 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
20369 intrinsic register names.
20370 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
20371 intrinsic register enum values and macros.
20373 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
20375 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
20376 for load/store addressing form.
20377 (nds32_print_operand_address): Likewise.
20379 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
20382 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
20383 based on LABEL_REF. Remove useless assertion.
20384 (pic_address_needs_scratch): Fix formatting.
20385 (sparc_legitimize_pic_address): Minor tweaks.
20386 (sparc_delegitimize_address): Adjust assertion accordingly.
20387 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
20388 into symbolic_operand.
20389 (movsi_high_pic_label_ref): Likewise.
20390 (movsi_lo_sum_pic_label_ref): Likewise.
20391 (movdi_pic_label_ref): Likewise.
20392 (movdi_high_pic_label_ref): Likewise.
20393 (movdi_lo_sum_pic_label_ref): Likewise.
20395 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
20397 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
20398 custom LIB_SPEC setup.
20400 2018-04-06 Ruslan Bukin <br@bsdpad.com>
20401 Kito Cheng <kito.cheng@gmail.com>
20403 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
20404 * config/riscv/freebsd.h: New.
20406 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
20408 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
20409 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
20412 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
20413 Kito Cheng <kito.cheng@gmail.com>
20415 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
20416 nds32_output_call, nds32_symbol_binds_local_p): New functions.
20417 * config/nds32/nds32-protos.h (nds32_output_call,
20418 nds32_output_return): Declare.
20419 * config/nds32/nds32.md: Refine all the call and return patterns.
20421 2018-04-06 Jakub Jelinek <jakub@redhat.com>
20424 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
20425 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
20427 PR rtl-optimization/84872
20428 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
20429 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
20430 EDGE_CROSSING edge.
20432 2018-04-06 Tamar Christina <tamar.christina@arm.com>
20434 * expr.c (copy_blkmode_to_reg): Revert 254862.
20435 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
20437 2018-04-06 Richard Biener <rguenther@suse.de>
20439 PR middle-end/85244
20440 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
20441 after seeing a component reference with an adjacent field. Treat
20442 refs to arrays at struct end of external decls similar to
20443 refs to unconstrained commons.
20445 2018-04-06 Jakub Jelinek <jakub@redhat.com>
20448 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
20449 look through SAVE_EXPRs with non-side-effects argument. Adjust
20451 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
20454 2018-04-06 Richard Biener <rguenther@suse.de>
20456 PR middle-end/85180
20457 * alias.c (find_base_term): New wrapper around find_base_term
20458 unwinding CSELIB_VAL_PTR changes.
20459 (find_base_term): Do not restore CSELIB_VAL_PTR during the
20462 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20464 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
20466 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
20467 constant definitions.
20468 ("nop"): lr 0,0 -> nopr r0
20469 ("nop_lr0", "nop_lr1"): New insn definitions.
20471 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
20473 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
20474 NDS32_V3PUSH_AVAILABLE_P macro.
20476 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
20477 Chung-Ju Wu <jasonwucj@gmail.com>
20479 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
20480 (nds32*-*-*): Add float and fpu_config into supported_defaults.
20481 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
20482 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
20483 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
20484 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
20485 * config/nds32/constraints.md: New constraints and checking for hard
20486 float configuration.
20487 * config/nds32/iterators.md: New mode iterator and attribute for hard
20488 float configuration.
20489 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
20491 * config/nds32/nds32-fpu.md: New file.
20492 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
20493 deal with hard float code generation.
20494 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
20496 (abi_type, float_reg_number): New enum type.
20497 * config/nds32/nds32-predicates.c: New predicates for hard float.
20498 * config/nds32/nds32-protos.h: Declare functions for hard float.
20499 * config/nds32/nds32.c: Implementation for hard float configuration.
20500 * config/nds32/nds32.h: Definitions for hard float configuration.
20501 * config/nds32/nds32.md: Include hard float machine description and
20502 modify patterns for hard float configuration.
20503 * config/nds32/nds32.opt: New options for hard float configuration.
20504 * config/nds32/predicates.md: New predicates for hard float
20507 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
20509 * common/config/nds32/nds32-common.c
20510 (nds32_option_optimization_table): Enable -mreleax-hint by default.
20512 2018-04-05 Jakub Jelinek <jakub@redhat.com>
20514 PR middle-end/85195
20515 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
20516 CONSTRUCTOR_ELT (ctor, ...)->value.
20518 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
20521 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
20523 2018-04-05 Tom de Vries <tom@codesourcery.com>
20526 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
20529 2018-04-05 Shiva Chen <shiva0217@gmail.com>
20530 Kito Cheng <kito.cheng@gmail.com>
20532 * config/nds32/constraints.md (U33): Fine-tune checking condition.
20533 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
20534 * config/nds32/nds32.h (nds32_16bit_address_type): Add
20535 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
20537 2018-04-05 Shiva Chen <shiva0217@gmail.com>
20538 Kito Cheng <kito.cheng@gmail.com>
20540 * config/nds32/constraints.md (Ufe): New memory constraint.
20541 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
20542 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
20543 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
20545 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
20546 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
20548 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20550 * config/nds32/nds32.md: Use optimize_size in the condition for
20551 alu-shift instructions.
20553 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20555 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
20557 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20559 * config/nds32/nds32.md (negsi2): Refine pattern.
20561 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
20562 Chung-Ju Wu <jasonwucj@gmail.com>
20564 * config/nds32/iterators.md (shift_rotate): New code iterator.
20565 (shift): New code attribute.
20566 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
20567 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
20568 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
20569 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
20570 bit-wise operations.
20571 (andsi3, *andsi3): Ditto.
20572 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
20573 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
20574 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
20575 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
20576 nds32_ior_operand, nds32_xor_operand): New predicates.
20578 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20580 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
20581 (addsi3, subsi3): ... this.
20583 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20585 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
20587 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20589 * config/nds32/nds32.md: Adjust indention.
20591 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
20593 * config/nds32/nds32.md (feature): New attribute.
20595 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20597 * config/nds32/nds32.md (subtype): New attribute.
20599 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20602 * config/arm/arm-builtins.c (arm_expand_builtin): Change
20603 expansion to perform a bitwise AND of the argument followed by a
20604 boolean negation of the result.
20606 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
20608 PR rtl-optimization/84878
20609 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
20610 the basic block. Assert the use reference is not artificial and that
20611 it has an associated insn.
20613 2018-04-04 Michael Matz <matz@suse.de>
20615 * builtins.c (compute_objsize): Pass correct operand
20616 to array_at_struct_end_p.
20618 2018-04-04 Richard Biener <rguenther@suse.de>
20621 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
20622 from contexts for DINFO_LEVEL_TERSE and below.
20624 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
20626 * config/nds32/nds32-doubleword.md (move_<mode>): Require
20627 resiter_operand condition.
20628 * config/nds32/nds32.md (*move<mode>): Ditto.
20630 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
20631 Monk Chiang <sh.chiang04@gmail.com>
20633 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
20635 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
20637 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
20639 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
20640 Kito Cheng <kito.cheng@gmail.com>
20642 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
20643 nds32_cond_code_str, output_cond_branch,
20644 output_cond_branch_compare_zero, nds32_expand_cbranch,
20645 nds32_expand_cstore, nds32_expand_movcc,
20646 nds32_output_cbranchsi4_equality_zero,
20647 nds32_output_cbranchsi4_equality_reg,
20648 nds32_output_cbranchsi4_equality_reg_or_const_int,
20649 nds32_output_cbranchsi4_greater_less_zero: New functions.
20650 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
20651 nds32_expand_cstore, nds32_expand_movcc,
20652 nds32_output_cbranchsi4_equality_zero,
20653 nds32_output_cbranchsi4_equality_reg,
20654 nds32_output_cbranchsi4_equality_reg_or_const_int,
20655 nds32_output_cbranchsi4_greater_less_zero): Declare.
20656 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
20657 nds32_rimm11s_operand): New predicates.
20658 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
20659 * config/nds32/nds32.md: Rewrite all the branch and conditional move
20662 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
20664 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
20665 * config/nds32/nds32.md: Ditto.
20666 * config/nds32/pipelines.md: Ditto.
20668 2018-04-04 Richard Biener <rguenther@suse.de>
20670 PR tree-optimization/85168
20671 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
20672 propagating abnormals.
20674 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
20676 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
20678 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
20679 Kito Cheng <kito.cheng@gmail.com>
20681 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
20682 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
20683 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
20684 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
20685 * config/nds32/nds32.md (sibcall_internal): New.
20686 (sibcall_register): Remove.
20687 (sibcall_immediate): Remove.
20688 (sibcall_value_internal): New.
20689 (sibcall_value_register): Remove.
20690 (sibcall_value_immediate): Remove.
20691 * config/nds32/predicates.md (nds32_general_register_operand): New.
20692 (nds32_call_address_operand): New.
20694 2018-04-03 Jakub Jelinek <jakub@redhat.com>
20696 PR rtl-optimization/85167
20697 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
20698 bb_defs if *split_p, instead preinitialize it to NULL.
20700 PR tree-optimization/85156
20701 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
20702 evaluating the argument multiple times.
20704 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
20706 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
20708 (_mm_cvtpd_ps): Likewise.
20709 (_mm_cvttpd_epi32): Likewise.
20710 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
20711 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
20712 vector, pixel, and bool following altivec.h include.
20714 2018-04-03 Martin Sebor <msebor@redhat.com>
20716 * doc/extend.texi (Common Function Attributes): Clarify.
20717 (const attribute): Likewise.
20718 (pure attribute): Likewise.
20720 2018-04-03 Jakub Jelinek <jakub@redhat.com>
20723 * config/i386/i386.c (ix86_expand_vector_set): Use
20724 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
20726 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
20728 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
20729 instructions when changing rounding bits to preserve precision bits
20730 in the x87 control word.
20732 2018-04-03 Martin Liska <mliska@suse.cz>
20734 PR tree-optimization/82491
20735 * rtl.h (strip_offset_and_add): Replace += suboffset with
20736 poly_uint64 () + suboffset.
20738 2018-03-29 Martin Liska <mliska@suse.cz>
20739 Martin Jambor <mjambor@suse.cz>
20742 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
20743 param_type is not an integral or pointer type.
20745 2018-04-03 Richard Biener <rguenther@suse.de>
20747 * sese.h (recompute_all_dominators): Remove.
20749 2018-04-02 Martin Sebor <msebor@redhat.com>
20751 * doc/invoke.texi (-Wrestrict): Fix typos.
20753 2018-04-02 Jim Wilson <jimw@sifive.com>
20755 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
20756 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
20757 (<optab>di3, <optab>si3_extend): Likewise.
20758 (<optab>si3_mask, <optab>si3_mask_1): New.
20759 (<optab>di3_mask, <optab>di3_mask_1): New.
20760 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
20761 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
20762 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
20764 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
20766 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
20769 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
20771 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
20772 (nds32_canonicalize_comparison): New function.
20774 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
20775 Kito Cheng <kito.cheng@gmail.com>
20776 Kuan-Lin Chen <kuanlinchentw@gmail.com>
20778 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
20779 * config/nds32/constants.md (unspec_volatile_element): Add
20780 UNSPEC_VOLATILE_RELAX_GROUP.
20781 * config/nds32/nds32-relax-opt.c: New file.
20782 * config/nds32/nds32-predicates.c
20783 (nds32_symbol_load_store_p): New function.
20784 * config/nds32/nds32-protos.h
20785 (nds32_symbol_load_store_p): Declare function.
20786 (make_pass_nds32_relax_opt): Declare new rtl pass function.
20787 * config/nds32/nds32.c
20788 (nds32_register_pass): New function to register pass.
20789 (nds32_register_passes): New function to register passes.
20790 * config/nds32/nds32.md (relax_group): New pattern.
20791 * config/nds32/nds32.opt (mrelax-hint): New option.
20792 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
20794 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
20796 * config/nds32/t-nds32: Modify files dependency.
20798 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
20800 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
20801 (PROFILE_HOOK): Define its implementation.
20803 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
20805 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
20806 type and 32-bit size.
20808 2018-04-01 Jakub Jelinek <jakub@redhat.com>
20810 PR middle-end/85090
20811 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
20812 (V_128_256): New mode iterator.
20813 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
20814 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
20815 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
20817 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
20818 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
20820 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
20823 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
20824 NaN inputs correctly.
20826 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
20829 * config/rs6000/vsx.md (??r): New mode attribute.
20830 (*vsx_mov<mode>_64bit): Use it.
20831 (*vsx_mov<mode>_32bit): Likewise.
20833 2018-03-30 Martin Sebor <msebor@redhat.com>
20835 PR tree-optimization/84818
20836 * builtins.c (check_access): Use warning_n.
20838 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20841 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
20843 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
20846 2018-03-30 Julia Koval <julia.koval@intel.com>
20849 * x86-tune.def (movx, partial_reg_dependency): Enable for
20852 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
20854 PR inline-asm/84985
20855 * lra-constraints.c (process_alt_operands): Move setting
20856 this_alternative_matches below.
20858 2018-03-29 Martin Liska <mliska@suse.cz>
20861 * doc/invoke.texi: Document how LTO works with debug info.
20862 Describe auto-load support of binutils. Mention 'x86-64'
20863 as valid option value of -march option.
20865 2018-03-29 Jakub Jelinek <jakub@redhat.com>
20867 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
20870 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
20871 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
20872 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
20875 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
20878 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
20879 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
20880 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
20881 for RS6000_BTM_POWERPC64.
20882 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
20883 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
20884 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
20887 (DIVDEU): Likewise.
20889 2018-03-28 Carl Love <cel@us.ibm.com>
20892 2017-09-27 Carl Love <cel@us.ibm.com>
20894 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
20895 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
20896 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
20899 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20901 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
20902 instead of __vector bool.
20903 (_mm_max_pu8): Likewise.
20904 (_mm_min_pi16): Likewise.
20906 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
20909 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
20910 (DIVWEUO): Likewise.
20911 (DIVDEO): Likewise.
20912 (DIVDEUO): Likewise.
20913 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
20914 DIVWEUO and DIVDEUO.
20915 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
20916 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
20917 (div_extend): Likewise.
20918 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
20920 (__builtin_divweuo): Likewise.
20921 (__builtin_divdeo): Likewise.
20922 (__builtin_divdeuo): Likewise.
20924 2018-03-28 Jakub Jelinek <jakub@redhat.com>
20927 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
20928 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
20930 PR tree-optimization/82004
20931 * gimple-match-head.c (optimize_pow_to_exp): New function.
20932 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
20933 Don't fold to exp if optimize_pow_to_exp is false.
20935 2018-03-28 Martin Liska <mliska@suse.cz>
20938 * calls.c (initialize_argument_information): Fix trailing space.
20939 * common.opt: Fix typo and provide better explanation for
20940 -fsanitize-coverage option.
20941 * config/i386/i386.opt: Fix typo.
20943 2018-03-28 Jakub Jelinek <jakub@redhat.com>
20944 Martin Liska <mliska@suse.cz>
20947 * gimplify.c (asan_poison_variable): Don't do the check for
20948 gimplify_omp_ctxp here.
20949 (gimplify_decl_expr): Do it here.
20950 (gimplify_target_expr): Likewise.
20952 2018-03-28 Martin Liska <mliska@suse.cz>
20955 * config/i386/i386.c (ix86_function_arg_advance): Do not call
20956 chkp_type_bounds_count if MPX is not enabled.
20958 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
20960 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
20962 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
20965 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
20966 function to create the function decl for complex long double
20967 multiply and divide for -mabi=ieeelongdouble.
20968 (init_float128_ieee): Call it.
20970 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
20973 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
20974 -fcf-protection=branch -mibt.
20975 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
20977 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20980 * config/arm/arm.c (arm_valid_symbolic_address): Handle
20981 arm_word_relocations.
20983 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
20986 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
20987 extern array declarations.
20989 2018-03-27 Richard Biener <rguenther@suse.de>
20991 PR middle-end/84067
20992 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
20993 explicit single_use checks.
20995 2018-03-27 Richard Biener <rguenther@suse.de>
20997 PR tree-optimization/85082
20998 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
21001 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21003 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
21004 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
21005 Turn on fasynchronous-unwind-tables and funwind-tables.
21007 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
21010 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
21011 (*bmi_blsr_<mode>_ccz): Ditto.
21013 2018-03-26 Tom de Vries <tom@codesourcery.com>
21015 PR tree-optimization/85063
21016 * omp-general.c (offloading_function_p): New function. Factor out
21018 * omp-offload.c (pass_omp_target_link::gate): ... here.
21019 * omp-general.h (offloading_function_p): Declare.
21020 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
21021 with attribute omp declare target for offloading functions.
21023 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
21025 PR tree-optimization/84005
21026 * tree-data-ref.h (get_base_for_alignment): Declare.
21027 * tree-data-ref.c (get_base_for_alignment_1): New function.
21028 (get_base_for_alignment): Likewise.
21029 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
21030 get_base_for_alignment to find a suitable base object, instead
21031 of always using drb->base_address.
21033 2018-03-23 Jakub Jelinek <jakub@redhat.com>
21035 PR inline-asm/85022
21036 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
21037 known size by default.
21039 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
21041 PR inline-asm/85030
21042 * lra-constraints.c (process_alt_operands): Don't match BLKmode
21043 and non BLKmode operands.
21045 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21048 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
21049 Clean up attributes.
21051 2018-03-23 Richard Biener <rguenther@suse.de>
21054 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
21055 we are going to emit early debug for LTO.
21057 2018-03-23 Jakub Jelinek <jakub@redhat.com>
21059 PR inline-asm/85034
21060 * function.c (match_asm_constraints_1): Don't optimize if input
21061 doesn't satisfy general_operand predicate for output's mode.
21063 PR inline-asm/85022
21064 * alias.c (write_dependence_p): Don't require for x_canonicalized
21065 non-VOIDmode if x has VOIDmode.
21068 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
21069 just don't try to optimize it rather than assert it never happens.
21071 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
21073 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
21074 macro expansions for definition of ST_INTERNAL_<mode> and
21075 LD_INTERNAL_<mode> builtins.
21076 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
21078 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
21080 (altivec_expand_st_builtin): Likewise.
21081 (altivec_expand_builtin): Remove calls to deleted functions.
21082 (rs6000_address_for_altivec): Delete this function.
21083 * config/rs6000/vector.md: Remove expands for
21084 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
21086 2018-03-22 Sudakshina Das <sudi.das@arm.com>
21089 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
21090 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
21091 re-computing once computed.
21092 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
21093 (arm_init_machine_status): Initialize
21094 machine->static_chain_stack_bytes.
21096 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
21099 * doc/extend.texi: Add four new prototypes for vec_ld.
21100 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
21101 definitions for more logical presentation.
21102 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
21103 entries for V1TI variants of __builtin_altivec_ld builtin.
21104 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
21105 handling of V1TI variant of LVX icode pattern.
21106 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
21107 (rs6000_gimple_fold_builtin): Likewise.
21108 (altivec_init_builtins): Add code to define
21109 __builtin_altivec_lvx_v1ti function.
21111 2018-03-22 Jakub Jelinek <jakub@redhat.com>
21113 PR inline-asm/84941
21114 * function.c (match_asm_constraints_1): Don't do the optimization
21115 if input isn't a REG, SUBREG, MEM or constant.
21117 2018-03-22 Tom de Vries <tom@codesourcery.com>
21119 PR tree-optimization/84956
21120 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
21121 bb_has_abnormal_pred.
21123 2018-03-22 Jakub Jelinek <jakub@redhat.com>
21126 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
21127 DECL_INITIAL (decl) to decl at the end.
21128 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
21129 adjust the comment.
21131 2018-03-21 Joseph Myers <joseph@codesourcery.com>
21133 * doc/extend.texi (__builtin_tgmath): Document when complex
21134 integer types are treated as _Complex _Float64.
21136 2018-03-21 Tom de Vries <tom@codesourcery.com>
21138 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
21140 2018-03-21 Jakub Jelinek <jakub@redhat.com>
21142 PR tree-optimization/84960
21143 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
21144 if it is ENTRY block, move them into single succ of ENTRY in that case.
21146 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
21148 PR tree-optimization/84811
21149 * poly-int.h (poly_span_traits): Remove the T3 parameter and
21150 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
21151 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
21152 (known_subrange_p): Update accordingly. Cast each value involved
21153 in the size comparison, rather than casting the result of the
21156 2018-03-21 Jakub Jelinek <jakub@redhat.com>
21158 PR tree-optimization/84982
21159 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
21160 by flipping the least significant bit rather than all bits from
21161 bitpos to bitpos + bitsize - 1.
21163 2018-03-21 Nathan Sidwell <nathan@acm.org>
21165 * doc/extend.texi (Deprecated Features): Remove mention of
21166 long-deleted deprecations.
21168 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21171 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
21172 * configure: Regenerate.
21174 2018-03-21 Tom de Vries <tom@codesourcery.com>
21176 PR tree-optimization/83126
21177 * tree-parloops.c (num_phis): New function.
21178 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
21180 2018-03-21 Nathan Sidwell <nathan@acm.org>
21182 * doc/extend.texi (Deprecated Features): Update deprecated flags,
21183 mention anon-struct/union members and trailing attributes.
21185 2018-03-21 Bin Cheng <bin.cheng@arm.com>
21187 PR tree-optimization/84969
21188 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
21189 builtin memset partitions if they set different rhs values.
21191 2018-03-21 Jakub Jelinek <jakub@redhat.com>
21193 PR rtl-optimization/84989
21194 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
21195 VEC_DUPLICATE with scalar result mode.
21197 2018-03-21 Martin Liska <mliska@suse.cz>
21200 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
21201 not intended return statement.
21203 2018-03-21 Martin Liska <mliska@suse.cz>
21206 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
21207 (chkp_find_bound_slots_1): Limit number of iterations.
21209 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
21212 * Minor grammar fixes for x86 options.
21214 2018-03-20 Jakub Jelinek <jakub@redhat.com>
21217 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
21218 holding REG_CFA_RESTORE notes, instead turn them into a USE.
21220 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
21223 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
21224 (altivec_lvx_<mode>_1op): Likewise.
21225 (altivec_stvx_<mode>_2op): Likewise.
21226 (altivec_stvx_<mode>_1op): Likewise.
21227 (altivec_lvx_<VM2:mode>): New define_expand.
21228 (altivec_stvx_<VM2:mode>): Likewise.
21229 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
21230 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
21231 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
21232 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
21233 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
21234 (rs6000_gen_lvx): Likewise.
21235 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
21236 (altivec_expand_stv_builtin): Likewise.
21237 (altivec_expand_builtin): Likewise.
21238 * config/rs6000/vector.md: Likewise.
21240 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21243 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
21246 2018-03-20 Richard Biener <rguenther@suse.de>
21249 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
21250 sign-conversions as zero, fall back to standard scalar_stmt
21253 2018-03-20 Martin Liska <mliska@suse.cz>
21256 * predict.c (rebuild_frequencies): Handle case when we have
21257 PROFILE_ABSENT, but flag_guess_branch_prob is false.
21259 2018-03-20 Jakub Jelinek <jakub@redhat.com>
21262 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
21263 flag_section_anchors.
21264 * varasm.c (use_blocks_for_decl_p): Remove hack for
21265 dw2_force_const_mem.
21268 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
21270 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
21271 be created, use lowpart_subreg of operands[0] rather than operands[0]
21273 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
21274 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
21275 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
21276 and n constraint instead of aarch64_shift_imm_di and Usd.
21277 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
21278 (*aarch64_<optab>_reg_minus<mode>3): ... this.
21280 2018-03-20 Sudakshina Das <sudi.das@arm.com>
21283 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
21284 to favor GPR over NEON registers.
21285 (<shift>di3_neon): Likewise.
21287 2018-03-20 Tom de Vries <tom@codesourcery.com>
21290 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
21291 (nvptx_process_pars): Emit bar.sync asap and alap.
21293 2018-03-20 Tom de Vries <tom@codesourcery.com>
21296 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
21297 seen_label if seen_label is already set.
21299 2018-03-20 Jakub Jelinek <jakub@redhat.com>
21302 * config/i386/i386.c (fold_builtin_cpu): For features above 31
21303 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
21304 Use 1U instead of 1. Formatting fixes.
21307 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
21308 instead of TREE_TYPE (s1) for the return value.
21310 2018-03-19 Jakub Jelinek <jakub@redhat.com>
21312 PR tree-optimization/84946
21313 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
21314 bitsize + bitsize in poly_uint64 rather than poly_int64.
21317 * dwarf2asm.c: Include fold-const.c.
21318 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
21319 of decl rather than decl itself.
21321 PR rtl-optimization/84643
21322 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
21324 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
21327 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
21328 calling assemble_variable.
21330 2018-03-19 Sudakshina Das <sudi.das@arm.com>
21333 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
21334 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
21336 2018-03-19 Jim Wilson <jimw@sifive.com>
21339 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
21340 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
21341 (riscv_first_stack_step): Likewise.
21342 (riscv_option_override): Use STACK_BOUNDARY instead of
21343 MIN_STACK_BOUNDARY.
21344 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
21345 MIN_STACK_BOUNDARY.
21346 (BIGGEST_ALIGNMENT): Set to 128.
21347 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
21348 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
21351 2018-03-19 Richard Biener <rguenther@suse.de>
21353 PR tree-optimization/84933
21354 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
21355 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
21357 2018-03-19 Richard Biener <rguenther@suse.de>
21359 PR tree-optimization/84859
21360 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
21361 (cond_if_else_store_replacement): Perform sinking operation on
21362 single-store BBs regardless of MAX_STORES_TO_SINK setting.
21363 Generalize what a BB with a single eligible store is.
21365 2018-03-19 Richard Biener <rguenther@suse.de>
21367 PR tree-optimization/84929
21368 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
21369 chrec_is_positive against non-chrec arg.
21371 2018-03-19 Tamar Christina <tamar.christina@arm.com>
21374 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
21376 2018-03-18 Martin Liska <mliska@suse.cz>
21378 PR rtl-optimization/84635
21379 * regrename.c (build_def_use): Use matches_mode only when
21382 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
21384 PR tree-optimization/84913
21385 * tree-vect-loop.c (vectorizable_reduction): Don't try to
21386 vectorize chains of COND_EXPRs.
21388 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
21390 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
21392 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
21394 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
21396 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
21398 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
21400 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
21401 Kito Cheng <kito.cheng@gmail.com>
21403 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
21404 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
21405 (nds32_adjust_reg_alloc_order): New function.
21406 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
21408 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
21410 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
21411 nds32_print_operand, nds32_print_operand_address): Use
21412 HOST_WIDE_INT_PRINT_DEC instead.
21414 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
21416 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
21418 2018-03-17 Jakub Jelinek <jakub@redhat.com>
21421 * config/i386/i386.c (initial_ix86_tune_features,
21422 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
21423 unsigned long long.
21424 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
21425 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
21426 rather than 1u << ix86_tune. Formatting fix.
21427 (ix86_option_override_internal): Change ix86_arch_mask from
21428 unsigned int to unsigned HOST_WIDE_INT, initialize to
21429 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
21430 (ix86_function_specific_restore): Likewise.
21432 2018-03-16 Jakub Jelinek <jakub@redhat.com>
21435 * postreload.c (reload_combine_recognize_pattern): Perform
21436 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
21437 truncate_int_for_mode the result for the destination's mode.
21440 * hsa-gen.c (mem_type_for_type): Fix comment typo.
21441 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
21443 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
21446 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
21449 * lra-assigns.c (lra_split_hard_reg_for): Don't use
21450 regno_allocno_class_array and sorted_pseudos.
21451 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
21452 insns where regno is used.
21454 2018-03-16 Martin Liska <mliska@suse.cz>
21457 * multiple_target.c (create_dispatcher_calls): Redirect
21458 reference in the symbol table.
21460 2018-03-16 Martin Liska <mliska@suse.cz>
21463 * multiple_target.c (create_dispatcher_calls): Redirect also
21466 2018-03-16 Jakub Jelinek <jakub@redhat.com>
21470 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
21471 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
21472 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
21474 2018-03-16 Julia Koval <julia.koval@intel.com>
21476 * doc/invoke.texi (Skylake Server): Add CLWB.
21477 Cannonlake): Remove CLWB.
21479 2018-03-16 Jakub Jelinek <jakub@redhat.com>
21481 PR tree-optimization/84841
21482 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
21484 (FLOAT_ONE_CONST_TYPE): Define.
21485 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
21486 (sort_by_operand_rank): Put entries with higher constant_type last
21487 rather than first to match comments.
21489 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
21491 * config/nios2/nios2.md (movsi_internal): Fix thinko in
21494 2018-03-15 Jakub Jelinek <jakub@redhat.com>
21497 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
21498 check and use address of target always.
21500 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
21503 * config/i386/i386.c (indirect_thunk_needed): Update comments.
21504 (indirect_thunk_bnd_needed): Likewise.
21505 (indirect_thunks_used): Likewise.
21506 (indirect_thunks_bnd_used): Likewise.
21507 (indirect_return_needed): New.
21508 (indirect_return_bnd_needed): Likewise.
21509 (output_indirect_thunk_function): Add a bool argument for
21511 (output_indirect_thunk_function): Don't generate alias for
21512 function return thunk.
21513 (ix86_code_end): Call output_indirect_thunk_function to generate
21514 function return thunks.
21515 (ix86_output_function_return): Set indirect_return_bnd_needed
21516 and indirect_return_needed instead of indirect_thunk_bnd_needed
21517 and indirect_thunk_needed.
21519 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
21521 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
21522 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
21523 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
21525 2018-03-15 David Malcolm <dmalcolm@redhat.com>
21526 Paul Hua <paul.hua.gm@gmail.com>
21529 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
21531 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
21533 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
21534 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
21535 resp. SFmode cases.
21537 2018-03-15 Tamar Christina <tamar.christina@arm.com>
21540 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
21541 instead of GET_MODE_SIZE when comparing Units.
21543 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
21546 * varasm.c (hash_section): Return an unchangeble hash value
21547 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
21548 Return !aarch64_can_use_per_function_literal_pools_p ().
21550 2018-03-15 Jakub Jelinek <jakub@redhat.com>
21553 * optabs.c (emit_conditional_move): Pass address of cmode's copy
21554 rather than address of cmode as last argument to prepare_cmp_insn.
21556 2018-03-15 Julia Koval <julia.koval@intel.com>
21558 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
21559 F_AVX512VNNI, F_AVX512BITALG): New.
21561 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
21564 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
21565 insn for floating-point loads and stores.
21567 2018-03-14 Carl Love <cel@us.ibm.com>
21569 * config/rs6000/rs6000-c.c: Add macro definitions for
21570 ALTIVEC_BUILTIN_VEC_PERMXOR.
21571 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
21572 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
21573 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
21574 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
21576 * config/doc/extend.texi: Add prototypes for vec_permxor.
21578 2018-03-14 David Malcolm <dmalcolm@redhat.com>
21581 * diagnostic-show-locus.c (class layout_point): Convert m_line
21582 from int to linenum_type.
21583 (line_span::comparator): Use linenum "compare" function when
21584 comparing line numbers.
21585 (test_line_span): New function.
21586 (layout_range::contains_point): Convert param "row" from int to
21588 (layout_range::intersects_line_p): Likewise.
21589 (layout::will_show_line_p): Likewise.
21590 (layout::print_source_line): Likewise.
21591 (layout::should_print_annotation_line_p): Likewise.
21592 (layout::print_annotation_line): Likewise.
21593 (layout::print_leading_fixits): Likewise.
21594 (layout::annotation_line_showed_range_p): Likewise.
21595 (struct line_corrections): Likewise for field m_row.
21596 (line_corrections::line_corrections): Likewise for param "row".
21597 (layout::print_trailing_fixits): Likewise.
21598 (layout::get_state_at_point): Likewise.
21599 (layout::get_x_bound_for_row): Likewise.
21600 (layout::print_line): Likewise.
21601 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
21602 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
21603 * input.c (selftest::test_linenum_comparisons): New function.
21604 (selftest::input_c_tests): Call it.
21605 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
21606 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
21607 * selftest.h (ASSERT_GT): New macro.
21608 (ASSERT_GT_AT): New macro.
21609 (ASSERT_LT): New macro.
21610 (ASSERT_LT_AT): New macro.
21612 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
21614 PR rtl-optimization/84780
21615 * combine.c (distribute_links): Don't make a link based on pc_rtx.
21617 2018-03-14 Martin Liska <mliska@suse.cz>
21619 * tree.c (record_node_allocation_statistics): Use
21620 get_stats_node_kind.
21621 (get_stats_node_kind): New function extracted from
21622 record_node_allocation_statistics.
21623 (free_node): Use get_stats_node_kind.
21625 2018-03-14 Richard Biener <rguenther@suse.de>
21627 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
21628 that the value-set of ANTIC_IN doesn't grow.
21631 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
21633 (BB_VISITED_WITH_VISITED_SUCCS): New define.
21634 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
21636 2018-03-14 Julia Koval <julia.koval@intel.com>
21638 * config.gcc (icelake-client, icelake-server): New.
21640 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
21641 (initial_ix86_arch_features): Ditto.
21642 (PTA_SKYLAKE): Add SGX.
21643 (PTA_ICELAKE): Remove.
21644 (PTA_ICELAKE_CLIENT): New.
21645 (PTA_ICELAKE_SERVER): New.
21646 (ix86_option_override_internal): Split up icelake on icelake client and
21648 (get_builtin_code_for_version): Ditto.
21649 (fold_builtin_cpu): Ditto.
21650 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
21651 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
21652 * config/i386/i386.h (processor_type): Ditto.
21653 * doc/invoke.texi: Ditto.
21655 2018-03-14 Jakub Jelinek <jakub@redhat.com>
21658 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
21659 INTEGER_CST offset, add it together with bitpos / 8 and
21660 sign extend based on POINTER_SIZE.
21664 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
21667 * config/i386/constraints.md (Yc): New register constraint.
21668 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
21669 Use Yc constraint for alternative 2 of operand 0. Remove
21670 preferred_for_speed attribute.
21672 2018-03-14 Richard Biener <rguenther@suse.de>
21674 PR tree-optimization/84830
21675 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
21676 with the old one to avoid oscillations.
21678 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
21681 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
21683 (assign_by_spills): Return a flag of reload assignment failure.
21684 Do not process the reload assignment failures. Do not spill other
21685 reload pseudos if they has the same reg class. Update n if
21687 (lra_assign): Add a return arg. Set up from the result of
21688 assign_by_spills call.
21689 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
21690 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
21691 usage_insns if it is not NULL.
21692 (spill_hard_reg_in_range): New function.
21693 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
21694 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
21695 function prototypes.
21696 (lra_assign): Change prototype.
21697 * lra.c (lra): Add code to deal with fails by splitting hard reg
21700 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
21702 * config/riscv/riscv.opt (mrelax): New option.
21703 * config/riscv/riscv.c (riscv_file_start): Emit ".option
21704 "norelax" when riscv_mrelax is disabled.
21705 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
21707 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21710 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
21711 reassociation for int modes.
21713 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
21715 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
21716 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
21718 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
21719 * config/aarch64/aarch64-sve.md
21720 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
21721 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
21722 (*extend<mode><Vwide>2): Rename to...
21723 (aarch64_sve_extend<mode><Vwide>2): ...this.
21724 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
21725 renaming the old pattern to...
21726 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
21728 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
21729 define_expand, renaming the old pattern to...
21730 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
21731 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
21732 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
21733 account when deciding which SVE instruction the optab should use.
21734 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
21736 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
21738 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
21739 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
21740 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
21741 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
21742 (tlsdesc_small_<mode>): Turn a define_expand and use
21743 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
21744 (tlsdesc_small_advsimd_<mode>): ...this.
21745 (tlsdesc_small_sve_<mode>): New pattern.
21747 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
21749 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
21750 (UNSPEC_UMUL_HIGHPART): New constants.
21751 (MUL_HIGHPART): New int iteraor.
21752 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
21753 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
21755 (*<su>mul<mode>3_highpart): New define_insn.
21757 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
21760 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
21763 2018-03-13 Martin Liska <mliska@suse.cz>
21766 * (sem_item_optimizer::sem_item_optimizer): Initialize new
21768 (sem_item_optimizer::~sem_item_optimizer): Release it.
21769 (sem_item_optimizer::merge_classes): Register variable aliases.
21770 (sem_item_optimizer::fixup_pt_set): New function.
21771 (sem_item_optimizer::fixup_points_to_sets): Likewise.
21772 * ipa-icf.h: Declare new variables and functions.
21774 2018-03-13 Jakub Jelinek <jakub@redhat.com>
21776 PR middle-end/84834
21777 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
21778 integer_pow2p@2 and test integer_pow2p in condition.
21779 (A < 0 ? C : 0): Similarly for @1.
21781 PR middle-end/84831
21782 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
21783 characters starting at p contain '\0' character, don't look beyond
21787 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
21788 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
21791 * reg-stack.c (change_stack): Change update_end var from int to
21792 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
21793 also call set_block_for_insn on the newly added insns and rescan.
21796 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
21797 on the last operand.
21800 * tree.c (stabilize_reference_1): Return save_expr (e) for
21801 STATEMENT_LIST even if it doesn't have side-effects.
21803 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
21805 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
21807 2018-03-12 Renlin Li <renlin.li@arm.com>
21809 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
21810 aarch64_output_scalar_simd_mov_immediate.
21812 2018-03-12 Martin Sebor <msebor@redhat.com>
21814 PR tree-optimization/83456
21815 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
21816 for perfectly overlapping calls to memcpy.
21817 (gimple_fold_builtin_memory_chk): Same.
21818 (gimple_fold_builtin_strcpy): Handle no-warning.
21819 (gimple_fold_builtin_stxcpy_chk): Same.
21820 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
21822 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
21824 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
21825 parameter. Use it for SFmode.
21826 (rs6000_function_arg_advance_1): Adjust.
21827 (rs6000_function_arg): Adjust.
21828 (rs6000_gimplify_va_arg): Pass false for that new parameter.
21830 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
21832 PR rtl-optimization/84169
21833 PR rtl-optimization/84780
21834 * combine.c (can_combine_p): Check for a 2-insn combination whether
21835 the destination register is used between the two insns, too.
21837 2018-03-12 Richard Biener <rguenther@suse.de>
21839 PR tree-optimization/84803
21840 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
21841 for refs DR analysis didn't process.
21843 2018-03-12 Richard Biener <rguenther@suse.de>
21845 PR tree-optimization/84777
21846 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
21847 force-vectorize loops ignore whether we are optimizing for size.
21849 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
21851 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
21852 (TARGET_MD_ASM_ADJUST): Define.
21854 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
21855 Kito Cheng <kito.cheng@gmail.com>
21856 Chung-Ju Wu <jasonwucj@gmail.com>
21858 * config/nds32/nds32.c (nds32_compute_stack_frame,
21859 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
21860 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
21861 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
21862 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
21863 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
21864 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
21865 * config/nds32/nds32.md (prologue, epilogue): Use macro
21866 NDS32_V3PUSH_AVAILABLE_P to do checking.
21868 2018-03-11 Jakub Jelinek <jakub@redhat.com>
21871 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
21872 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
21873 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
21874 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
21875 addition of most attributes on !orig_type_die or the attribute not
21876 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
21878 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
21879 Chung-Ju Wu <jasonwucj@gmail.com>
21881 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
21882 __NDS32_VH__ macro.
21883 * config/nds32/nds32.opt (mvh): New option.
21885 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
21886 Chung-Ju Wu <jasonwucj@gmail.com>
21888 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
21890 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
21891 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
21894 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
21895 Chung-Ju Wu <jasonwucj@gmail.com>
21897 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
21899 * config/nds32/nds32-multiple.md (strlensi): New pattern.
21900 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
21902 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
21903 Kito Cheng <kito.cheng@gmail.com>
21904 Chung-Ju Wu <jasonwucj@gmail.com>
21906 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
21907 UNSPEC_FFMISM and UNSPEC_FLMISM.
21908 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
21909 for ffb, ffmism and flmism.
21910 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
21911 (unspec_ffmism): Ditto.
21912 (unspec_flmism): Ditto.
21913 (nds32_expand_builtin_impl): Check if string extension is available.
21914 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
21915 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
21917 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
21920 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
21923 * lra-assigns.c (assign_by_spills): Return a flag of reload
21924 assignment failure. Do not process the reload assignment
21925 failures. Do not spill other reload pseudos if they has the same
21927 (lra_assign): Add a return arg. Set up from the result of
21928 assign_by_spills call.
21929 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
21930 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
21931 usage_insns if it is not NULL.
21932 (spill_hard_reg_in_range): New function.
21933 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
21934 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
21935 function prototypes.
21936 (lra_assign): Change prototype.
21937 * lra.c (lra): Add code to deal with fails by splitting hard reg
21940 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
21943 * config/i386/i386.opt: Replace Enforcment with Enforcement.
21945 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
21948 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
21949 (dw_val_node): Add val_symbolic_view.
21950 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
21951 (symview_upper_bound): New.
21952 (new_line_info_table): Initialize symviews_since_reset.
21953 (dwarf2out_source_line): Count symviews_since_reset and set
21954 symview_upper_bound.
21955 (dw_val_equal_p): Handle symview.
21956 (add_AT_symview): New.
21957 (print_dw_val): Handle symview.
21958 (attr_checksum, attr_checksum_ordered): Likewise.
21959 (same_dw_val_p, size_of_die): Likewise.
21960 (value_format, output_die): Likewise.
21961 (add_high_low_attributes): Use add_AT_symview for entry_view.
21962 (dwarf2out_finish): Reset symview_upper_bound, clear
21965 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
21968 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
21969 Add strict argument and use it.
21970 (rs6000_split_multireg_move): Update for new strict argument.
21971 (mem_operand_gpr): Disallow all non-offsettable addresses.
21972 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
21974 2018-03-09 Jakub Jelinek <jakub@redhat.com>
21977 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
21978 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
21979 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
21982 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
21983 decl, use remap_type if we want to use the type.
21985 2018-03-09 Martin Sebor <msebor@redhat.com>
21987 PR tree-optimization/84526
21988 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
21990 (builtin_access::generic_overlap): Be prepared to handle non-array
21993 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
21995 PR rtl-optimization/84682
21996 * lra-constraints.c (process_address_1): Check is_address flag
21997 for address constraints.
21998 (process_alt_operands): Likewise.
21999 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
22000 preprocess_constraints.
22001 * recog.h (preprocess_constraints): Add oploc parameter.
22003 * recog.c (preprocess_constraints): Test address_operand for
22004 CT_ADDRESS constraints.
22006 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
22009 * lra-assigns.c (assign_by_spills): Return a flag of reload
22010 assignment failure. Do not process the reload assignment
22011 failures. Do not spill other reload pseudos if they has the same
22013 (lra_assign): Add a return arg. Set up from the result of
22014 assign_by_spills call.
22015 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
22016 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
22017 usage_insns if it is not NULL.
22018 (spill_hard_reg_in_range): New function.
22019 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
22020 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
22021 function prototypes.
22022 (lra_assign): Change prototype.
22023 * lra.c (lra): Add code to deal with fails by splitting hard reg
22026 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22029 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
22030 Accept complain bool parameter. Only emit errors if it is true.
22031 (arm_parse_cpu_option_name): Likewise.
22032 (arm_target_thumb_only): Adjust callers of the above.
22033 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
22034 prototype to take a default true bool parameter.
22035 (arm_parse_arch_option_name): Likewise.
22037 2018-03-09 David Malcolm <dmalcolm@redhat.com>
22038 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
22042 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
22043 * configure: Regenerate.
22044 * configure.ac ("linker --version-script option"): New.
22045 ("linker soname option"): New.
22047 2018-03-09 Richard Biener <rguenther@suse.de>
22049 PR tree-optimization/84775
22050 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
22051 immediate uses of predicate stmts and mark them modified.
22054 PR tree-optimization/84178
22055 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
22057 (version_loop_for_if_conversion): Delay update_ssa call.
22058 (tree_if_conversion): Delay update_ssa until after predicate
22061 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
22064 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
22065 when the function accesses prior frames.
22067 2018-03-08 Jakub Jelinek <jakub@redhat.com>
22070 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
22071 gen_llsym, otherwise call maybe_gen_llsym.
22073 PR inline-asm/84742
22074 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
22075 has ',' character inside of it.
22077 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22080 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
22081 as clobbering CC_REGNUM.
22083 2018-03-08 Richard Biener <rguenther@suse.de>
22085 PR middle-end/84552
22086 * tree-scalar-evolution.c: Include tree-into-ssa.h.
22087 (follow_copies_to_constant): Do not follow SSA names registered
22090 2018-03-08 Richard Biener <rguenther@suse.de>
22092 PR tree-optimization/84178
22093 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
22095 (version_loop_for_if_conversion): Delay update_ssa call.
22096 (tree_if_conversion): Delay update_ssa until after predicate
22099 2018-03-08 David Malcolm <dmalcolm@redhat.com>
22101 PR tree-optimization/84178
22102 * tree-if-conv.c (release_bb_predicate): Remove the
22103 the assertion that the stmts have NULL use_ops.
22104 Discard the statements, asserting that they haven't
22105 yet been added to a BB.
22107 2018-03-08 Richard Biener <rguenther@suse.de>
22109 PR tree-optimization/84746
22110 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
22111 (phi_translate): Pass in destination ANTIC_OUT set.
22112 (phi_translate_1): Likewise. For a simplified result lookup
22113 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
22114 (phi_translate_set): Adjust.
22115 (do_pre_regular_insertion): Likewise.
22116 (do_pre_partial_partial_insertion): Likewise.
22118 2018-03-08 Martin Liska <mliska@suse.cz>
22120 PR gcov-profile/84735
22121 * doc/gcov.texi: Document usage of profile files.
22122 * gcov-io.h: Document changes in the format.
22124 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
22128 * dwarf2out.c (struct dw_line_info_table): Update comments for
22130 (FORCE_RESET_NEXT_VIEW): New.
22131 (FORCE_RESETTING_VIEW_P): New.
22132 (RESETTING_VIEW_P): Check for -1 too.
22133 (ZERO_VIEW_P): Likewise.
22134 (new_line_info_table): Force-reset next view.
22135 (dwarf2out_begin_function): Likewise.
22136 (dwarf2out_source_line): Simplify zero_view_p initialization.
22137 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
22138 view directly. Omit view when omitting .loc at line 0.
22140 2018-03-08 Jakub Jelinek <jakub@redhat.com>
22142 PR tree-optimization/84740
22143 * tree-switch-conversion.c (process_switch): Call build_constructors
22144 only if info.phi_count is non-zero.
22146 PR tree-optimization/84739
22147 * tree-tailcall.c (find_tail_calls): Check call arguments against
22148 DECL_ARGUMENTS (current_function_decl) rather than
22149 DECL_ARGUMENTS (func) when checking for tail recursion.
22151 2018-03-07 Jakub Jelinek <jakub@redhat.com>
22153 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
22154 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
22155 Volker Reichelt's entry and add entries for people that perform
22156 GCC fuzzy testing and report numerous bugs.
22158 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
22161 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
22162 readonly data in sdata, if that is disabled.
22163 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
22164 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
22165 -mreadonly-in-sdata option.
22167 2018-03-07 Martin Sebor <msebor@redhat.com>
22169 PR tree-optimization/84468
22170 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
22171 basic block when looking for nul assignment.
22173 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
22176 * except.h (output_function_exception_table): Adjust prototype.
22177 * except.c (output_function_exception_table): Remove FNNAME parameter
22178 and add SECTION parameter. Ouput one part of the table at a time.
22179 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
22180 the first part of the exception table and emit unwind directives.
22181 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
22182 (i386_pe_seh_cold_init): Likewise.
22183 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
22184 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
22185 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
22186 (ix86_output_call_insn): Emit a nop in one more case for SEH.
22187 * config/i386/winnt.c: Include except.h.
22188 (struct seh_frame_state): Add reg_offset, after_prologue and
22189 in_cold_section fields.
22190 (i386_pe_seh_end_prologue): Set seh->after_prologue.
22191 (i386_pe_seh_cold_init): New function.
22192 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
22193 to seh->in_cold_section.
22194 (seh_emit_push): Record the offset of the push.
22195 (seh_emit_save): Record the offet of the save.
22196 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
22197 Test seh->after_prologue to disregard the epilogue.
22198 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
22199 (i386_pe_end_cold_function): New function.
22201 2018-03-07 Jakub Jelinek <jakub@redhat.com>
22204 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
22205 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
22208 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
22210 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
22211 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
22213 PR middle-end/84723
22214 * multiple_target.c: Include tree-inline.h and intl.h.
22215 (expand_target_clones): Diagnose and fail if node->definition and
22216 !tree_versionable_function_p (node->decl).
22218 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
22220 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
22222 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
22223 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
22224 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
22226 2018-03-06 Jakub Jelinek <jakub@redhat.com>
22229 * combine.c (try_combine): Use reg_or_subregno instead of handling
22230 just paradoxical SUBREGs and REGs.
22232 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
22234 * config/arc/arc.c (arc_finalize_pic): Remove function.
22235 (arc_must_save_register): We use single base PIC register, remove
22236 checks to save/restore the PIC register.
22237 (arc_expand_prologue): Likewise.
22238 * config/arc/arc-protos.h (arc_set_default_type_attributes):
22240 (arc_verify_short): Likewise.
22241 (arc_attr_type): Likewise.
22242 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
22243 (walk_stores): Likewise.
22244 (arc_address_cost): Make it static.
22245 (arc_verify_short): Likewise.
22246 (branch_dest): Likewise.
22247 (arc_attr_type): Likewise.
22248 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
22249 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
22250 (arc_final_prescan_insn): Remove inserting the nops due to
22251 hardware hazards. It is done in reorg step.
22252 (insn_length_variant_t): Remove.
22253 (insn_length_parameters_t): Likewise.
22254 (arc_insn_length_parameters): Likewise.
22255 (arc_get_insn_variants): Likewise.
22256 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
22258 2018-03-06 Jakub Jelinek <jakub@redhat.com>
22260 PR inline-asm/84683
22261 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
22264 PR tree-optimization/84687
22265 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
22267 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
22269 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22271 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
22272 Rename to ppc_speculation_barrier.
22273 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
22274 __builtin_ppc_speculation_barrier.
22276 2018-03-05 Jakub Jelinek <jakub@redhat.com>
22279 * combine.c (combine_simplify_rtx): Don't try to simplify if
22280 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
22283 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
22285 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
22286 to 32 bytes when compiling for POWER9.
22288 2018-03-05 Jakub Jelinek <jakub@redhat.com>
22291 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
22292 regparm >= 3 with no arg reg available also for calls with
22293 flag_force_indirect_call. Pass decl to ix86_function_regparm.
22296 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
22298 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
22300 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
22303 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
22305 2018-03-05 Richard Biener <rguenther@suse.de>
22307 PR tree-optimization/84486
22308 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
22309 When inserting a __builtin_assume_aligned call set the LHS
22310 SSA name alignment info accordingly.
22312 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
22314 PR tree-optimization/84114
22315 * config/aarch64/aarch64.c (aarch64_reassociation_width)
22316 Avoid reassociation of FLOAT_MODE addition.
22318 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
22320 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
22321 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
22322 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
22323 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
22324 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
22325 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
22326 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
22328 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
22329 __builtin_ia32_wbinvd): New builtins.
22330 (SPECIAL_ARGS2): New.
22331 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
22332 (SPECIAL_ARGS2): New.
22333 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
22334 (ix86_valid_target_attribute_inner_p): Ditto.
22335 (ix86_init_mmx_sse_builtins): Add special_args2.
22336 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
22337 TARGET_WBNOINVD_P): New.
22338 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
22339 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
22340 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
22341 * config/i386/immintrin.h (_wbinvd): New intrinsic.
22342 * config/i386/pconfigintrin.h: New file.
22343 * config/i386/wbnoinvdintrin.h: Ditto.
22344 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
22346 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
22348 2018-03-05 Richard Biener <rguenther@suse.de>
22350 PR tree-optimization/84670
22351 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
22353 (BB_VISITED_WITH_VISITED_SUCCS): New define.
22354 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
22355 (compute_antic_aux): Only assert the number of values in ANTIC_IN
22356 doesn't grow if all successors (recursively) were visited at least
22359 2018-03-05 Richard Biener <rguenther@suse.de>
22361 PR tree-optimization/84650
22362 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
22363 if executed in the loop pipeline.
22365 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
22367 * doc/configfiles.texi (Configuration Files): Move info about
22368 conditionalizing $target-protos.h to...
22369 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
22370 differs from $target-protos.h.
22372 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
22373 Chung-Ju Wu <jasonwucj@gmail.com>
22375 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
22376 * config/nds32/nds32-multiple.md (setmemsi): Define.
22377 * config/nds32/nds32-memory-manipulation.c
22378 (nds32_gen_dup_4_byte_to_word_value): New.
22379 (emit_setmem_word_loop): New.
22380 (emit_setmem_byte_loop): New.
22381 (nds32_expand_setmem_loop): New.
22382 (nds32_expand_setmem_loop_v3m): New.
22383 (nds32_expand_setmem_unroll): New.
22384 (nds32_expand_setmem): New.
22386 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
22387 Chung-Ju Wu <jasonwucj@gmail.com>
22389 * config/nds32/nds32-memory-manipulation.c
22390 (nds32_emit_load_store): New.
22391 (nds32_emit_post_inc_load_store): New.
22392 (nds32_emit_mem_move): New.
22393 (nds32_emit_mem_move_block): New.
22394 (nds32_expand_movmemsi_loop_unknown_size): New.
22395 (nds32_expand_movmemsi_loop_known_size): New.
22396 (nds32_expand_movmemsi_loop): New.
22397 (nds32_expand_movmemsi_unroll): New.
22398 (nds32_expand_movmemqi): Rename ...
22399 (nds32_expand_movmemsi): ... to this.
22400 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
22401 (movmemsi): ... to this.
22402 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
22403 (nds32_expand_movmemsi): ... to this.
22405 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
22406 Monk Chiang <sh.chiang04@gmail.com>
22407 Chung-Ju Wu <jasonwucj@gmail.com>
22409 * config/nds32/nds32-protos.h
22410 (nds32_expand_load_multiple): New arguments.
22411 (nds32_expand_store_multiple): Ditto.
22412 (nds32_valid_multiple_load_store): Rename ...
22413 (nds32_valid_multiple_load_store_p): ... to this.
22414 * config/nds32/nds32-memory-manipulation.c
22415 (nds32_expand_load_multiple): Refine implementation.
22416 (nds32_expand_store_multiple): Ditto.
22417 * config/nds32/nds32-multiple.md
22418 (load_multiple): Update nds32_expand_load_multiple interface.
22419 (store_multiple): Update nds32_expand_store_multiple interface.
22420 * config/nds32/nds32-predicates.c
22421 (nds32_valid_multiple_load_store): Rename ...
22422 (nds32_valid_multiple_load_store_p): ... to this and refine
22424 * config/nds32/predicates.md
22425 (nds32_load_multiple_and_update_address_operation): New predicate.
22426 (nds32_store_multiple_and_update_address_operation): New predicate.
22428 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
22429 Chung-Ju Wu <jasonwucj@gmail.com>
22431 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
22432 (combo): New attribute.
22433 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
22435 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
22437 * config/nds32/nds32.opt: Change -mcmodel= default value.
22439 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
22440 Monk Chiang <sh.chiang04@gmail.com>
22441 Chung-Ju Wu <jasonwucj@gmail.com>
22443 * config/nds32/constants.md (unspec_element): New enum.
22444 * config/nds32/constraints.md (Umw): New constraint.
22445 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
22446 * config/nds32/nds32-intrinsic.md: Likewise.
22447 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
22448 (nds32_valid_smw_lwm_base_p): New.
22449 (nds32_output_smw_single_word): New.
22450 (nds32_output_lmw_single_word): New.
22451 (nds32_expand_unaligned_load): New.
22452 (nds32_expand_unaligned_store): New.
22453 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
22454 (nds32_output_smw_single_word): Declare.
22455 (nds32_output_lmw_single_word): Declare.
22456 (nds32_expand_unaligned_load): Declare.
22457 (nds32_expand_unaligned_store): Declare.
22458 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
22459 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
22460 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
22461 NDS32_BUILTIN_UASTORE_DW.
22462 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
22465 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
22466 Kito Cheng <kito.cheng@gmail.com>
22467 Chung-Ju Wu <jasonwucj@gmail.com>
22469 * config/nds32/nds32-intrinsic.c
22470 (nds32_expand_builtin_null_ftype_reg): Delete.
22471 (nds32_expand_builtin_reg_ftype_imm): Ditto.
22472 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
22473 (nds32_read_argument): New.
22474 (nds32_legitimize_target): Ditto.
22475 (nds32_legitimize_argument): Ditto.
22476 (nds32_check_constant_argument): Ditto.
22477 (nds32_expand_unop_builtin): Ditto.
22478 (nds32_expand_unopimm_builtin): Ditto.
22479 (nds32_expand_binop_builtin): Ditto.
22480 (nds32_builtin_decl_impl): Ditto.
22481 (builtin_description): Ditto.
22482 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
22483 (nds32_init_builtins_impl): Ditto.
22484 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
22485 (nds32_builtin_decl): New.
22486 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
22487 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
22489 2018-03-02 Jeff Law <law@redhat.com>
22491 * reorg.c (stop_search_p): Handle DEBUG_INSN.
22492 (redundant_insn, fill_simple_delay_slots): Likewise.
22493 (fill_slots_from_thread): Likewise.
22494 * resource.c (mark_referenced_resources): Likewise.
22495 (mark_set_resources, find_dead_or_set_registers): Likewise.
22497 2018-03-02 Jakub Jelinek <jakub@redhat.com>
22499 * substring-locations.h (format_warning_va): Formatting fix for
22500 ATTRIBUTE_GCC_DIAG.
22501 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
22503 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
22504 * substring-locations.c: Include intl.h.
22505 (format_warning_va): Turned into small wrapper around
22506 format_warning_n_va, renamed to ...
22507 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
22508 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
22510 (format_warning_at_substring_n): New function.
22511 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
22512 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
22513 format_warning_at_substring with just a shorter name instead of
22514 const function pointer.
22515 (fmtwarn_n): New function.
22516 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
22517 appropriate, get rid of all the fmtstr temporaries, move conditionals
22518 with G_() wrapped string literals directly into fmtwarn arguments,
22519 cast dir.len to (int), formatting fixes.
22521 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
22523 * doc/invoke.texi: Remove "Cilk Plus" references.
22525 2018-03-02 Jakub Jelinek <jakub@redhat.com>
22526 Richard Biener <rguenther@suse.de>
22529 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
22530 for error or warning attributes if CALL_FROM_THUNK_P is set.
22533 2018-03-02 Jakub Jelinek <jakub@redhat.com>
22536 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
22537 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
22540 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
22541 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
22543 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
22544 instead of -1U in last predictors element's probability member.
22546 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
22549 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
22550 arguments if they are comparable.
22552 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
22554 PR tree-optimization/84634
22555 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
22556 masks and masked_loop_p with a single loop_masks, making sure it's
22557 null for bb vectorization.
22559 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
22561 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
22562 (vect_analyze_data_ref_access): Use loop->safe_len rather than
22563 loop->force_vectorize to check whether there is no alias.
22565 2018-03-02 Jakub Jelinek <jakub@redhat.com>
22568 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
22570 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
22572 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
22573 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
22574 instead of a loop around prev_real_insn.
22575 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
22578 PR inline-asm/84625
22579 * config/i386/i386.c (ix86_print_operand): Use conditional
22580 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
22583 2018-03-02 Richard Biener <rguenther@suse.de>
22585 PR tree-optimization/84427
22586 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
22587 (bitmap_set_subtract_values): Rewrite to handle multiple
22590 (prune_clobbered_mems): Likewise.
22591 (phi_translate): Take edge instead of pred/phiblock.
22592 (phi_translate_1): Likewise.
22593 (phi_translate_set): Likewise. Insert all translated
22594 exprs for a value into the set, keeping possibly multiple
22595 expressions per value.
22596 (compute_antic_aux): Adjust for phi_translate changes.
22597 When intersecting union the expressions and prune those
22598 not in the final value set, keeping possibly multiple
22599 expressions per value. Do not use value-insertion
22600 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
22601 all expressions. Add verification that the value-sets
22602 only shrink during iteration.
22603 (compute_partial_antic_aux): Adjust for the phi_translate changes.
22604 (do_pre_regular_insertion): Likewise.
22605 (do_pre_partial_partial_insertion): Likewise.
22607 2018-03-02 Richard Biener <rguenther@suse.de>
22610 * config/darwin.c (saved_debug_info_level): New static global.
22611 (darwin_asm_lto_start): Disable debug info generation for LTO out.
22612 (darwin_asm_lto_end): Restore debug info generation settings.
22614 2018-03-01 Martin Liska <mliska@suse.cz>
22617 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
22618 volatile arguments.
22620 2018-03-01 Richard Biener <rguenther@suse.de>
22623 * dwarf2out.c (gen_variable_die): Properly handle late VLA
22624 type annotation with LTO when debug was disabled at compile-time.
22626 2018-03-01 Matthew Fortune <mfortune@gmail.com>
22628 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
22630 (mips_final_postscan_insn): Likewise.
22632 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
22634 PR rtl-optimization/84528
22635 * alias.c (init_alias_target): Add commentary.
22636 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
22637 a unique base value if the frame pointer is not eliminated
22638 to the stack pointer.
22640 2018-03-01 Tom de Vries <tom@codesourcery.com>
22642 PR rtl-optimization/83327
22643 * lra-int.h (hard_regs_spilled_into): Declare.
22644 * lra.c (hard_regs_spilled_into): Define.
22645 (init_reg_info): Init hard_regs_spilled_into.
22646 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
22647 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
22648 (process_bb_lives): Handle hard_regs_spilled_into.
22649 (lra_create_live_ranges_1): Before doing liveness propagation, clear
22650 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
22652 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
22654 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
22655 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
22656 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
22657 * config/rs6000/aix72.h: New file.
22659 2018-02-28 Jakub Jelinek <jakub@redhat.com>
22661 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
22662 instead of warning_at with conditional singular and plural messages
22666 * stor-layout.c (update_alignment_for_field): For
22667 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
22668 && !DECL_PACKED (field), do the alignment update, just use
22669 only desired_align instead of MAX (type_align, desired_align)
22671 (place_field): Don't do known_align < desired_align handling
22672 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
22673 is non-NULL, instead do it after rli->prev_field handling and
22674 only if not within a bitfield word. For DECL_PACKED (field)
22675 use type_align of BITS_PER_UNIT.
22677 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
22679 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
22680 superfluous parentheses and trailing spaces.
22682 2018-02-28 Richard Biener <rguenther@suse.de>
22684 PR tree-optimization/84584
22685 * graphite-scop-detection.c (scop_detection::add_scop): Discard
22686 SCoPs with fake exit edge.
22688 2018-02-28 Martin Liska <mliska@suse.cz>
22691 * timevar.c (timer::print): Fix format to properly print 100%
22694 2018-02-28 Richard Biener <rguenther@suse.de>
22696 PR middle-end/84607
22697 * genmatch.c (capture_info::walk_match): Do not mark
22698 captured expressions without operands as expr_p given
22699 they act more like predicates and should be subject to
22700 "lost tail" side-effect preserving.
22702 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
22704 PR rtl-optimization/81611
22705 * auto-inc-dec.c (attempt_change): Move dead note from
22706 mem_insn if it's the next use of regno
22707 (find_address): Take address use of reg holding
22708 non-incremented value. Add parm to limit search to the named
22710 (merge_in_block): Attempt to use a mem insn that is the next
22711 use of the original regno.
22713 2018-02-27 Martin Sebor <msebor@redhat.com>
22716 * doc/invoke.texi (-Wmissing-attributes): New option.
22717 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
22719 2018-02-27 Martin Sebor <msebor@redhat.com>
22721 PR translation/84207
22722 * diagnostic-core.h (warning_n, error_n, inform_n): Change
22723 n argument to unsigned HOST_WIDE_INT.
22724 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
22725 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
22726 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
22727 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
22729 2018-02-27 Richard Biener <rguenther@suse.de>
22731 PR tree-optimization/84512
22732 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
22733 Do not use the estimate returned from record_stmt_cost for
22734 the scalar iteration cost but sum properly using add_stmt_cost.
22736 2018-02-27 Richard Biener <rguenther@suse.de>
22738 PR tree-optimization/84466
22739 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
22740 Adjust last change to less strictly validate use operands.
22742 2018-02-27 Martin Liska <mliska@suse.cz>
22744 PR gcov-profile/84548
22745 * gcov.c (process_file): Allow partial overlap and consider it
22746 also as group functions.
22747 (output_lines): Properly calculate range of lines for a group.
22749 2018-02-27 Martin Liska <mliska@suse.cz>
22751 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
22752 'ggc' suffixes. Change first column width.
22753 (timer::print): Fix formatting of the column.
22755 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
22757 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
22758 preserve inline entry blocks for the sake of debug inline
22759 entry point markers alone.
22760 (remove_unused_locals): Suggest in comments a better place to
22761 force the preservation of inline entry blocks that are
22762 otherwise unused, but do not preserve them.
22764 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
22766 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
22768 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
22771 * config/i386/constraints.md (Bs): Replace
22772 ix86_indirect_branch_register with
22773 TARGET_INDIRECT_BRANCH_REGISTER.
22775 * config/i386/i386.md (indirect_jump): Likewise.
22776 (tablejump): Likewise.
22777 (*sibcall_memory): Likewise.
22778 (*sibcall_value_memory): Likewise.
22779 Peepholes of indirect call and jump via memory: Likewise.
22780 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
22781 (*sibcall_value_GOT_32): Likewise.
22782 * config/i386/predicates.md (indirect_branch_operand): Likewise.
22783 (GOT_memory_operand): Likewise.
22784 (call_insn_operand): Likewise.
22785 (sibcall_insn_operand): Likewise.
22786 (GOT32_symbol_operand): Likewise.
22787 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
22789 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
22791 PR rtl-optimization/83496
22792 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
22793 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
22794 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
22795 redundant insn, if any.
22796 (relax_delay_slots): Likewise.
22797 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
22799 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
22801 PR tree-optimization/83965
22802 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
22803 that grouped statements are part of a reduction chain. Return
22804 true if the statement is not marked as a reduction itself but
22805 is part of a group.
22806 (vect_recog_dot_prod_pattern): Don't check whether the statement
22807 is part of a group here.
22808 (vect_recog_sad_pattern): Likewise.
22809 (vect_recog_widen_sum_pattern): Likewise.
22811 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
22814 * final.c (rest_of_clean_state): Also look for calls inside sequences.
22816 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
22819 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
22821 (ix86_output_indirect_function_return): New prototype.
22822 (ix86_split_simple_return_pop_internal): Likewise.
22823 * config/i386/i386.c (indirect_return_via_cx): New.
22824 (indirect_return_via_cx_bnd): Likewise.
22825 (indirect_thunk_name): Handle return va CX_REG.
22826 (output_indirect_thunk_function): Create alias for
22827 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
22828 (ix86_output_indirect_jmp): Remove the bool argument.
22829 (ix86_output_indirect_function_return): New function.
22830 (ix86_split_simple_return_pop_internal): Likewise.
22831 * config/i386/i386.md (*indirect_jump): Don't pass false
22832 to ix86_output_indirect_jmp.
22833 (*tablejump_1): Likewise.
22834 (simple_return_pop_internal): Change it to define_insn_and_split.
22835 Call ix86_split_simple_return_pop_internal to split it for
22836 -mfunction-return=.
22837 (simple_return_indirect_internal): Call
22838 ix86_output_indirect_function_return instead of
22839 ix86_output_indirect_jmp.
22841 2018-02-26 Jakub Jelinek <jakub@redhat.com>
22844 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
22845 memset and value initialization afterwards.
22847 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
22849 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
22851 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22854 * common/config/aarch64/aarch64-common.c
22855 (aarch_option_optimization_table[]): Switch
22856 off fomit-frame-pointer
22858 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
22859 Chung-Ju Wu <jasonwucj@gmail.com>
22861 * config/nds32/nds32-multiple.md (load_multiple): Disallow
22863 (store_multiple): Ditto.
22865 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
22867 * config.gcc: Add --with-cpu support for nds32 target.
22868 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
22869 * config/nds32/nds32.opt: Add -mcpu= option.
22871 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
22873 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
22874 isel=yes): Warn for these deprecated options.
22876 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
22878 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
22879 ISA_2_5_MASKS_EMBEDDED.
22881 2018-02-23 Jakub Jelinek <jakub@redhat.com>
22883 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
22884 p->max as pointers rather than using iterative_hash_expr.
22886 2018-02-23 Carl Love <cel@us.ibm.com>
22888 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
22889 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
22891 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
22892 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
22893 P8V_BUILTIN_VEC_VUNSIGNED2.
22895 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
22898 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
22899 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
22901 * lra-constraints.c (curr_insn_transform): Set up
22902 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
22904 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
22905 * lra-eliminations.c (spill_pseudos): Ditto.
22906 (process_insn_for_elimination): Ditto.
22907 * lra-lives.c (reg_early_clobber_p): Use the new macros.
22908 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
22909 LRA_NON_CLOBBERED_ALT.
22911 2018-02-22 Martin Sebor <msebor@redhat.com>
22913 PR tree-optimization/84480
22914 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
22915 to maybe_diag_stxncpy_trunc. Call it.
22916 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
22917 from gimple_fold_builtin_strcpy. Print inlining stack.
22918 (handle_builtin_stxncpy): Print inlining stack.
22919 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
22921 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
22924 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
22925 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
22926 and -fcheck-pointer-bounds are used together.
22927 (indirect_thunk_prefix): New enum.
22928 (indirect_thunk_need_prefix): New function.
22929 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
22930 "_nt" instead of "_bnd" for NOTRACK prefix.
22931 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
22932 (output_indirect_thunk_function): Likewise.
22934 (ix86_code_end): Update output_indirect_thunk_function calls.
22935 (ix86_output_indirect_branch_via_reg): Replace
22936 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
22937 (ix86_output_indirect_branch_via_push): Likewise.
22938 (ix86_output_function_return): Likewise.
22939 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
22940 incompatible with -fcf-protection=branch and
22941 -fcheck-pointer-bounds.
22943 2018-02-22 Steve Ellcey <sellcey@cavium.com>
22946 * config/aarch64/aarch64.c (aarch64_print_address_internal):
22947 Change gcc_assert call to output_operand_lossage.
22949 2018-02-22 Steve Ellcey <sellcey@cavium.com>
22951 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
22953 2018-02-22 DJ Delorie <dj@redhat.com>
22954 Sebastian Perta <sebastian.perta@renesas.com>
22955 Oleg Endo <olegendo@gcc.gnu.org>
22957 * config/rx/rx.c (rx_rtx_costs): New function.
22958 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
22960 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
22962 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
22964 2018-02-22 Martin Liska <mliska@suse.cz>
22967 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
22968 Add "native" as a possible value.
22970 2018-02-22 Martin Liska <mliska@suse.cz>
22973 * config/i386/i386.c (ix86_option_override_internal):
22974 Add "native" as a possible value for -march and -mtune.
22976 2018-02-22 Jakub Jelinek <jakub@redhat.com>
22979 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
22980 to all type variants.
22982 PR tree-optimization/84503
22983 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
22984 width as info->bitpos + info->bitsize - start.
22985 (merged_store_group::merge_overlapping): Simplify width computation.
22986 (check_no_overlap): New function.
22987 (imm_store_chain_info::try_coalesce_bswap): Compute expected
22988 start + width and last_order of the group, fail if check_no_overlap
22990 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
22991 to group if check_no_overlap fails.
22993 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
22995 * config/rs6000/altivec.md: Delete contraint arguments to
22996 define_expand, define_split, and define_peephole2, and in
22997 define_insn_and_split if always unused.
22998 * config/rs6000/darwin.md: Ditto.
22999 * config/rs6000/dfp.md: Ditto.
23000 * config/rs6000/rs6000.md: Ditto.
23001 * config/rs6000/sync.md: Ditto.
23002 * config/rs6000/vector.md: Ditto.
23003 * config/rs6000/vsx.md: Ditto.
23005 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
23007 * config/rs6000/altivec.md: Write output control strings as braced
23008 blocks instead of double-quoted strings.
23009 * config/rs6000/darwin.md: Ditto.
23010 * config/rs6000/rs6000.md: Ditto.
23011 * config/rs6000/vector.md: Ditto.
23012 * config/rs6000/vsx.md: Ditto.
23014 2018-02-21 Jason Merrill <jason@redhat.com>
23016 PR c++/84314 - ICE with templates and fastcall attribute.
23017 * attribs.c (build_type_attribute_qual_variant): Remove assert.
23019 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
23021 * ipa-cp.c (determine_versionability): Fix comment typos.
23023 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
23026 * ipa-cp.c (determine_versionability): Do not version functions caling
23029 2018-02-21 Martin Liska <mliska@suse.cz>
23032 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
23033 Add "native" as a possible value.
23034 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
23035 the macro when native cpu detection is available.
23037 2018-02-21 Martin Liska <mliska@suse.cz>
23040 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
23041 Add "native" as a possible value.
23042 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
23043 when native cpu detection is available.
23045 2018-02-21 Jakub Jelinek <jakub@redhat.com>
23046 Martin Sebor <msebor@redhat.com>
23048 PR tree-optimization/84478
23049 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
23051 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
23052 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
23053 support which is conservatively correct, for 2 only stay conservative
23054 for maxlen. Formatting and comment capitalization fixes. Add STRICT
23055 argument to the 2 argument get_range_strlen, adjust 6 arg
23056 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
23058 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
23059 (gimple_fold_builtin_strlen): Pass true as last argument to
23062 2018-02-20 Martin Sebor <msebor@redhat.com>
23064 PR middle-end/84095
23065 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
23066 (builtin_memref::set_base_and_offset): Same. Handle inner references.
23067 (builtin_memref::builtin_memref): Factor out parts into
23068 set_base_and_offset and call it.
23070 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
23072 PR middle-end/84406
23073 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
23074 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
23075 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
23076 search at the associated MODE_INT.
23078 2018-02-20 Jeff Law <law@redhat.com>
23080 PR middle-end/82123
23081 PR tree-optimization/81592
23082 PR middle-end/79257
23083 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
23084 for range data rather than using global data.
23085 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
23086 range data rather than using global data.
23087 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
23088 pass it to children as needed.
23089 (struct directive::fmtresult): Similarly.
23090 (struct directive::set_width): Similarly.
23091 (struct directive::set_precision): Similarly.
23092 (format_integer, format_directive, parse_directive): Similarly.
23093 (format_none): Accept unnamed vr_values parameter.
23094 (format_percent, format_floating, format_character): Similarly.
23095 (format_string, format_plain): Similarly.
23096 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
23097 the EVRP range analyzer for range data rather than using global data.
23098 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
23099 gimple-ssa-evrp-analyze.h
23100 (class sprintf_dom_walker): Add after_dom_children member function.
23101 Add evrp_range_analyzer member.
23102 (sprintf_dom_walker::before_dom_children): Call into the EVRP
23103 range analyzer as needed.
23104 (sprintf_dom_walker::after_dom_children): New member function.
23105 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
23107 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
23108 (evrp_range_analyzer::pop_to_marker): Likewise.
23110 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
23112 PR tree-optimization/84419
23113 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
23114 with the required type if its current type is compatible but
23117 2018-02-20 Jakub Jelinek <jakub@redhat.com>
23119 PR middle-end/82004
23120 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
23121 after vectorization.
23123 2018-02-20 Martin Liska <mliska@suse.cz>
23126 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
23127 possible values if we don't have a hint.
23129 2018-02-20 Martin Liska <mliska@suse.cz>
23133 * final.c (shorten_branches): Build align_tab array with one
23135 * opts.c (finish_options): Add alignment option limit check.
23136 (MAX_CODE_ALIGN): Likewise.
23137 (MAX_CODE_ALIGN_VALUE): Likewise.
23138 * doc/invoke.texi: Document maximum allowed option value for
23139 all -falign-* options.
23141 2018-02-19 Jakub Jelinek <jakub@redhat.com>
23144 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
23145 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
23146 * var-tracking.c (emit_note_insn_var_location): Remove all references
23147 to NOTE_INSN_CALL_ARG_LOCATION.
23148 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
23149 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
23150 Use copy_rtx_if_shared.
23151 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
23152 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
23153 (dwarf2out_var_location): Remove handling of
23154 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
23156 * final.c (final_scan_insn): Remove all references to
23157 NOTE_INSN_CALL_ARG_LOCATION.
23158 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
23159 before dumping final insns.
23160 * except.c (emit_note_eh_region_end): Remove all references to
23161 NOTE_INSN_CALL_ARG_LOCATION.
23162 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
23163 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
23164 * config/arc/arc.c (hwloop_optimize): Likewise.
23165 * config/arm/arm.c (create_fix_barrier): Likewise.
23166 * config/s390/s390.c (s390_chunkify_start): Likewise.
23167 * config/sh/sh.c (find_barrier): Likewise.
23168 * config/i386/i386.c (rest_of_insert_endbranch,
23169 ix86_seh_fixup_eh_fallthru): Likewise.
23170 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
23171 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
23172 * config/frv/frv.c (frv_function_prologue): Likewise.
23173 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
23175 (note_outside_basic_block_p): Remove all references to
23176 NOTE_INSN_CALL_ARG_LOCATION.
23177 * gengtype.c (adjust_field_rtx_def): Likewise.
23178 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
23180 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
23181 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
23184 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
23187 PR tree-optimization/84452
23188 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
23189 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
23192 2018-02-19 Martin Liska <mliska@suse.cz>
23195 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
23197 2018-02-19 Martin Liska <mliska@suse.cz>
23198 Richard Sandiford <richard.sandiford@linaro.org>
23200 PR tree-optimization/82491
23201 * gimple-fold.c (get_base_constructor): Make earlier bail out
23204 2018-02-19 Carl Love <cel@us.ibm.com>
23206 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
23207 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
23209 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
23210 P8V_BUILTIN_VEC_NEG.
23212 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
23214 * config/rl78/rl78.md (movdf): New define expand.
23216 2018-02-19 Martin Liska <mliska@suse.cz>
23219 * doc/invoke.texi: Fix typo.
23220 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
23222 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
23224 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
23225 handle rs6000_single_float and rs6000_double_float specially for
23228 2018-02-16 Jeff Law <law@redhat.com>
23230 * config/rx/rx.c (add_pop_cfi_notes): New function.;
23231 (pop_regs): Use it.
23233 2018-02-16 Jakub Jelinek <jakub@redhat.com>
23236 * ipa-inline.c (inline_small_functions): Fix a typo.
23238 2018-02-16 Nathan Sidwell <nathan@acm.org>
23240 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
23242 2018-02-16 Carl Love <cel@us.ibm.com>
23244 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
23245 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
23246 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
23247 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
23248 expansion to P8V_BUILTIN_VEC_FLOAT2.
23250 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
23252 PR rtl-optimization/70023
23253 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
23254 src_regno into account.
23256 2018-02-16 Carl Love <cel@us.ibm.com>
23258 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
23259 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
23260 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
23261 * config/rs6000/rs6000.c: Remove case statements for
23262 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
23263 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
23264 and P9V_BUILTIN_VEC_VINSERT4B.
23265 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
23266 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
23267 * config/rs6000/vsx.md:
23268 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
23271 2018-02-16 Carl Love <cel@us.ibm.com>
23273 * config/rs6000/altivec.h: Add builtin names vec_extract4b
23275 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
23277 * config/rs6000/rs6000-c.c: Add the definitions for
23278 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
23279 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
23280 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
23281 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
23282 definition for insert4b and define insn *insert3b_internal.
23283 * doc/extend.texi: Add documentation for vec_extract4b.
23285 2018-02-16 Nathan Sidwell <nathan@acm.org>
23287 * doc/extend.texi (Backwards Compatibility): Mention friend
23288 injection. Note for-scope is deprecated.
23289 * doc/invoke.texi (-ffriend-injection): Deprecate.
23291 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
23293 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
23294 that moved to I2, also allow destinations that are a paradoxical
23295 subreg (instead of a normal reg).
23297 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
23300 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
23303 2018-02-16 Richard Biener <rguenther@suse.de>
23305 PR tree-optimization/84037
23306 PR tree-optimization/84016
23308 * config/i386/i386.c (ix86_builtin_vectorization_cost):
23309 Adjust vec_construct for the fact we need additional higher latency
23310 128bit inserts for AVX256 and AVX512 vector builds.
23311 (ix86_add_stmt_cost): Scale vector construction cost for
23314 2018-02-16 Richard Biener <rguenther@suse.de>
23316 PR tree-optimization/84417
23317 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
23318 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
23319 (non_rewritable_lvalue_p): Likewise, use poly-ints.
23321 2018-02-16 Martin Liska <mliska@suse.cz>
23324 * internal-fn.def (ASAN_CHECK): Set proper flags.
23325 (ASAN_MARK): Likewise.
23327 2018-02-16 Julia Koval <julia.koval@intel.com>
23329 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
23330 from PTA_CANNONLAKE.
23332 2018-02-16 Jakub Jelinek <jakub@redhat.com>
23335 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
23336 Use ++iter rather than iter++ for std::list iterators.
23337 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
23338 defer deleting them until all nodes in the forest are processed. Do
23339 free even leaf nodes. Change to_process into auto_vec.
23342 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
23343 * vec.h (vec_default_construct): Use memset instead of placement new
23344 if BROKEN_VALUE_INITIALIZATION is defined.
23345 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
23346 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
23349 PR rtl-optimization/83723
23350 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
23351 * lra.c (lra_substitute_pseudo): Likewise. If true, use
23352 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
23354 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
23356 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
23358 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
23360 PR rtl-optimization/81443
23361 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
23362 from inner REGs to paradoxical SUBREGs.
23364 2018-02-16 Richard Biener <rguenther@suse.de>
23366 PR tree-optimization/84399
23367 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
23368 For operands we can analyze at their definition make sure we can
23369 analyze them at each use as well.
23371 2018-02-16 Richard Biener <rguenther@suse.de>
23373 PR tree-optimization/84190
23374 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
23375 volatile accesses if the decl isn't volatile.
23377 2018-02-15 Jason Merrill <jason@redhat.com>
23379 PR c++/84314 - ICE with templates and fastcall attribute.
23380 * attribs.c (build_type_attribute_qual_variant): Don't clobber
23381 TYPE_CANONICAL on an existing type.
23383 2018-02-15 Jakub Jelinek <jakub@redhat.com>
23385 PR tree-optimization/84383
23386 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
23387 dstoff nor call operand_equal_p if dstbase is NULL.
23389 PR tree-optimization/84334
23390 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
23391 also a CONSTANT_CLASS_P, punt.
23393 2018-02-14 Jim Wilson <jimw@sifive.com>
23395 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
23396 first SMALL_OPERAND check. New local min_second_step. Move assert
23397 to where locals are set. Add TARGET_RVC support.
23398 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
23400 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
23402 * doc/invoke.texi: Correct -Wformat-overflow code sample.
23404 2018-02-14 Martin Sebor <msebor@redhat.com>
23406 PR tree-optimization/83698
23407 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
23408 arrays constrain the offset range to their bounds.
23409 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
23410 (builtin_access::overlap): Avoid setting the size of overlap if it's
23412 (maybe_diag_overlap): Also consider arrays when deciding what values
23413 of offsets to include in diagnostics.
23415 2018-02-14 Martin Sebor <msebor@redhat.com>
23418 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
23419 that correspond to the kind of a declaration.
23421 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
23424 * config/pa/pa.md: Load address of PIC label using the linkage table
23425 if the label is nonlocal.
23427 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23429 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
23430 warning message if user requests -maltivec=be.
23431 * doc/invoke.texi: Document deprecation of -maltivec=be.
23433 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
23436 * config/rs6000/rs6000-c.c: Update definitions for
23437 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
23438 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
23440 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
23443 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
23444 add _get_ssp intrinsics. Remove argument from
23445 __builtin_ia32_rdssp[d|q].
23446 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
23447 * config/i386/i386-builtin.def: Remove argument from
23448 __builtin_ia32_rdssp[d|q].
23449 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
23450 ix86_expand_special_args_builtin for _rdssp[d|q].
23451 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
23452 Clear register before usage.
23453 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
23454 Add documentation for new _get_ssp and _inc_ssp intrinsics.
23456 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
23458 PR tree-optimization/84357
23459 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
23460 operand 1 of an ARRAY_REF too.
23462 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
23465 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
23466 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
23468 (set_of_reg): New struct.
23469 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
23470 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
23471 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
23473 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
23474 Split into bitclr, bitset, bitinvert patterns if appropriate.
23475 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
23476 use rx_fuse_in_memory_bitop.
23477 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
23478 to named insn, correct maximum insn length.
23480 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
23483 * machmode.def: Define a complex mode for PARTIAL_INT.
23484 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
23486 * doc/rtl.texi: Document CSPImode.
23487 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
23489 (msp430_hard_regno_nregs_with_padding): Likewise.
23491 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
23494 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
23496 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
23498 PR rtl-optimization/84169
23499 * combine.c (try_combine): New variable split_i2i3. Set it to true if
23500 we generated a parallel as new i3 and we split that to new i2 and i3
23501 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
23502 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
23503 those to i2, not i1. Partially rewrite this scan code.
23505 2018-02-13 Jakub Jelinek <jakub@redhat.com>
23508 * stor-layout.c (place_field): For variable length fields, adjust
23509 offset_align afterwards not just based on the field's alignment,
23510 but also on the size.
23512 PR middle-end/84309
23513 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
23514 of exps and logs in the use_exp2 case.
23516 2018-02-13 Jeff Law <law@redhat.com>
23518 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
23519 entry for "vector".
23521 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
23524 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
23528 * common.opt (gas-loc-support, gas-locview-support): New.
23529 (ginline-points, ginternal-reset-location-views): New.
23530 * doc/invoke.texi: Document them. Use @itemx where intended.
23531 (gvariable-location-views): Adjust.
23532 * target.def (reset_location_view): New.
23533 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
23534 (TARGET_RESET_LOCATION_VIEW): New.
23535 * doc/tm.texi: Rebuilt.
23536 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
23537 (dwarf2out_default_as_locview_support): New.
23538 (output_asm_line_debug_info): Use option variables.
23539 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
23540 (output_loc_list): Likewise.
23541 (add_high_low_attributes): Check option variables.
23542 Don't output entry view attribute in strict mode.
23543 (gen_inlined_subroutine_die): Check option variables.
23544 (dwarf2out_inline_entry): Likewise.
23545 (init_sections_and_labels): Likewise.
23546 (dwarf2out_early_finish): Likewise.
23547 (maybe_reset_location_view): New, from...
23548 (dwarf2out_var_location): ... here. Call it.
23549 * debug.h (dwarf2out_default_as_loc_support): Declare.
23550 (dwarf2out_default_as_locview_support): Declare.
23551 * hooks.c (hook_int_rtx_insn_0): New.
23552 * hooks.h (hook_int_rtx_insn_0): Declare.
23553 * toplev.c (process_options): Take -gas-loc-support and
23554 -gas-locview-support from dwarf2out. Enable
23555 -gvariable-location-views by default only with locview
23556 assembler support. Enable -ginternal-reset-location-views by
23557 default only if the target defines the corresponding hook.
23558 Enable -ginline-points by default if location views are
23559 enabled; force it disabled if statement frontiers are
23561 * tree-inline.c (expand_call_inline): Check option variables.
23562 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
23564 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
23566 PR tree-optimization/84321
23567 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
23568 handling. Also check whether the anti-range contains any values
23569 that satisfy the mask; switch to a VR_RANGE if not.
23571 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
23574 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
23576 2018-02-13 Martin Jambor <mjambor@suse.cz>
23579 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
23580 of call statements, also set location of a load to a temporary.
23582 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
23584 * config/rl78/rl78.c (add_vector_labels): New function.
23585 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
23586 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
23587 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
23588 which checks that no arguments are passed.
23589 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
23590 * doc/extend.texi: Documentation for the new attribute.
23592 2018-02-13 Andreas Schwab <schwab@suse.de>
23594 * config/riscv/linux.h (CPP_SPEC): Define.
23596 2018-02-13 Jakub Jelinek <jakub@redhat.com>
23599 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
23600 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
23601 OPTION_MASK_ISA_AES as first argument to def_builtin_const
23602 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
23603 instead of OPTION_MASK_ISA_PCLMUL as first argument to
23604 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
23605 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
23606 temporarily for AES and PCLMUL builtins.
23608 PR tree-optimization/84339
23609 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
23610 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
23613 PR middle-end/84309
23614 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
23615 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
23616 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
23618 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
23620 * omp-simd-clone.h: New file.
23621 * omp-simd-clone.c: Include omp-simd-clone.h.
23622 (expand_simd_clones): No longer static.
23623 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
23624 cgraph.h and omp-simd-clone.h.
23625 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
23626 (vect_recog_widen_shift_pattern): Formatting fix.
23627 (vect_pattern_recog_1): Don't check optab for calls.
23630 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
23631 operands[2] into a REG before using gen_lowpart on it.
23633 2018-02-12 Jeff Law <law@redhat.com>
23636 * config/sh/sh.c (find_barrier): Consider a sibling call
23639 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
23640 successfully back substituting a reg.
23642 2018-02-12 Richard Biener <rguenther@suse.de>
23644 PR tree-optimization/84037
23645 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
23646 parameter, move visited init to caller.
23647 (vect_slp_analyze_operations): Separate cost from validity
23648 check, initialize visited once for all instances.
23649 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
23651 * tree-vect-stmts.c (vect_model_simple_cost): Make early
23653 (vect_model_promotion_demotion_cost): Likewise.
23654 (vectorizable_bswap): Guard cost modeling with !slp_node
23655 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
23657 (vectorizable_call): Likewise.
23658 (vectorizable_conversion): Likewise.
23659 (vectorizable_assignment): Likewise.
23660 (vectorizable_shift): Likewise.
23661 (vectorizable_operation): Likewise.
23662 (vectorizable_store): Likewise.
23663 (vectorizable_load): Likewise.
23664 (vectorizable_condition): Likewise.
23665 (vectorizable_comparison): Likewise.
23667 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
23670 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
23671 (ASAN_MARK): Fix fnspec to account for return value, change pointer
23672 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
23674 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
23676 PR middle-end/83665
23677 * params.def (inline-min-speedup): Increase from 8 to 15.
23678 (max-inline-insns-auto): Decrease from 40 to 30.
23679 * ipa-split.c (consider_split): Add some buffer for function to
23680 be considered inlining candidate.
23681 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
23684 2018-02-12 Richard Biener <rguenther@suse.de>
23686 PR tree-optimization/84037
23687 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
23688 matched stmts if we cannot swap the non-matched ones.
23690 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
23692 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
23693 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
23694 _mm_maskz_scalef_round_ss): New intrinsics.
23695 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
23696 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
23697 __builtin_ia32_scalefss_round): Remove.
23698 (__builtin_ia32_scalefsd_mask_round,
23699 __builtin_ia32_scalefss_mask_round): New intrinsics.
23700 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
23701 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
23702 ((match_operand:VF_128 2 "<round_nimm_predicate>"
23703 "<round_constraint>")): Changed to ...
23704 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
23705 "<round_scalar_constraint>")): ... this.
23706 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
23707 %0, %1, %2<round_op3>}"): Changed to ...
23708 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
23709 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
23710 %2<round_scalar_mask_op3>}"): ... this.
23711 * config/i386/subst.md (round_scalar_nimm_predicate): New.
23713 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
23715 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
23716 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
23717 (_mm_maskz_sqrt_round_ss): New intrinsics.
23718 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
23719 (__builtin_ia32_sqrtsd_mask_round)
23720 (__builtin_ia32_sqrtss_mask_round): New builtins.
23721 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
23722 (__builtin_ia32_sqrtss_round): Remove.
23723 (__builtin_ia32_sqrtsd_mask_round)
23724 (__builtin_ia32_sqrtss_mask_round): New builtins.
23725 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
23726 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
23727 ((match_operand:VF_128 1 "vector_operand"
23728 "xBm,<round_constraint>")): Changed to ...
23729 ((match_operand:VF_128 1 "vector_operand"
23730 "xBm,<round_scalar_constraint>")): ... this.
23731 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
23732 %0, %2, %<iptr>1<round_op3>}): Changed to ...
23733 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
23734 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
23735 %<iptr>1<round_scalar_mask_op3>}): ... this.
23736 ((set_attr "prefix" "<round_prefix>")): Changed to ...
23737 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
23739 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
23742 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
23743 Cast vec_cmpeq result to correct type.
23744 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
23745 Cast vec_cmpgt result to correct type.
23747 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
23749 * final.c (final_scan_insn_1): Renamed from...
23750 (final_scan_insn): ... this. New wrapper, to recover
23751 seen from the outermost call in recursive ones.
23752 * config/sparc/sparc.c (output_return): Drop seen from call.
23753 (output_sibcall): Likewise.
23754 * config/visium/visium.c (output_branch): Likewise.
23756 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
23758 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
23761 2018-02-10 Alan Modra <amodra@gmail.com>
23764 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
23765 Specify LR as an input.
23767 2018-02-10 Jakub Jelinek <jakub@redhat.com>
23770 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
23771 remove_member_access_dummy_vars): New functions.
23772 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
23773 lower_omp_1, execute_lower_omp): Use them.
23775 PR rtl-optimization/84308
23776 * shrink-wrap.c (spread_components): Release todo vector.
23778 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
23780 PR rtl-optimization/57193
23781 * ira-color.c (struct allocno_color_data): Add member
23782 conflict_allocno_hard_prefs.
23783 (update_conflict_allocno_hard_prefs): New.
23784 (bucket_allocno_compare_func): Add a preference based on
23785 conflict_allocno_hard_prefs.
23786 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
23787 (color_allocnos): Remove a dead code. Initiate
23788 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
23790 2018-02-09 Jakub Jelinek <jakub@redhat.com>
23793 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
23794 constraint from =wa to wa. Avoid a subreg on the output operand,
23795 instead use a pseudo and subreg it in a move.
23796 (p9_xxbrd_<mode>): Changed to ...
23797 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
23798 (p9_xxbrd_v2df): New expander.
23799 (p9_xxbrw_<mode>): Changed to ...
23800 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
23801 (p9_xxbrw_v4sf): New expander.
23803 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
23805 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
23807 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
23810 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
23811 multiply in 32-bit mode.
23812 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
23813 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
23816 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
23818 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
23819 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
23820 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
23821 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
23823 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
23826 * dwarf2out.c (is_trivial_indirect_ref): New function.
23827 (dwarf2out_late_global_decl): Do not generate a location
23828 attribute for variables that have a non-trivial DECL_VALUE_EXPR
23829 and that are not defined in the current unit.
23831 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
23833 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
23834 instead of a libcall for UNORDERED.
23836 2018-02-09 Tamar Christina <tamar.christina@arm.com>
23839 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
23840 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
23842 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23845 * config/s390/s390.c (s390_set_current_function): Invoke
23846 s390_indirect_branch_settings also if fndecl didn't change.
23848 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
23850 * config/rs6000/rs6000.md (blockage): Set length to zero.
23852 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
23854 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
23856 2018-02-09 Jakub Jelinek <jakub@redhat.com>
23859 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
23860 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
23864 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
23865 PARALLEL incoming that failed vt_get_decl_and_offset check.
23867 PR middle-end/84237
23868 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
23869 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
23871 (get_variable_section): For decls in named .bss* sections pass true as
23872 second argument to bss_initializer_p.
23874 2018-02-09 Marek Polacek <polacek@redhat.com>
23875 Jakub Jelinek <jakub@redhat.com>
23878 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
23879 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
23880 Sync some changes from cxx_fold_indirect_ref.
23882 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
23884 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
23886 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
23887 (BLOCK_INLINE_ENTRY_LABEL): New.
23888 (dwarf2out_var_location): Disregard inline entry markers.
23889 (inline_entry_data): New struct.
23890 (inline_entry_data_hasher): New hashtable type.
23891 (inline_entry_data_hasher::hash): New.
23892 (inline_entry_data_hasher::equal): New.
23893 (inline_entry_data_table): New variable.
23894 (add_high_low_attributes): Add DW_AT_entry_pc and
23895 DW_AT_GNU_entry_view attributes if a pending entry is found
23896 in inline_entry_data_table. Add old entry_pc attribute only
23897 if debug nonbinding markers are disabled.
23898 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
23899 markers are enabled.
23900 (block_within_block_p, dwarf2out_inline_entry): New.
23901 (dwarf2out_finish): Check that no entries remained in
23902 inline_entry_data_table.
23903 * final.c (reemit_insn_block_notes): Handle inline entry notes.
23904 (final_scan_insn, notice_source_line): Likewise.
23905 (rest_of_clean_state): Skip inline entry markers.
23906 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
23908 * gimple.c (gimple_build_debug_inline_entry): New.
23909 * gimple.h (enum gimple_debug_subcode): Add
23910 GIMPLE_DEBUG_INLINE_ENTRY.
23911 (gimple_build_debug_inline_entry): Declare.
23912 (gimple_debug_inline_entry_p): New.
23913 (gimple_debug_nonbind_marker_p): Adjust.
23914 * insn-notes.def (INLINE_ENTRY): New.
23915 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
23916 inline entry marker notes.
23917 (print_insn): Likewise.
23918 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
23919 (INSN_DEBUG_MARKER_KIND): Likewise.
23920 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
23921 * tree-inline.c (expand_call_inline): Build and insert
23922 debug_inline_entry stmt.
23923 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
23924 inline entry blocks early, if nonbind markers are enabled.
23925 (dump_scope_block): Dump fragment info.
23926 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
23927 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
23928 (gimple_build_debug_inline_entry): New.
23929 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
23930 Enable/disable inline entry points too.
23931 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
23932 (DEBUG_INSN): Describe inline entry markers.
23934 * common.opt (gvariable-location-views): New.
23935 (gvariable-location-views=incompat5): New.
23936 * config.in: Rebuilt.
23937 * configure: Rebuilt.
23938 * configure.ac: Test assembler for view support.
23939 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
23940 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
23941 * dwarf2out.c (var_loc_view): New typedef.
23942 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
23943 (dwarf2out_locviews_in_attribute): New.
23944 (dwarf2out_locviews_in_loclist): New.
23945 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
23946 (enum dw_line_info_opcode): Add LI_adv_address.
23947 (struct dw_line_info_table): Add view.
23948 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
23949 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
23950 (zero_view_p): New variable.
23951 (ZERO_VIEW_P): New macro.
23952 (output_asm_line_debug_info): New.
23953 (struct var_loc_node): Add view.
23954 (add_AT_view_list, AT_loc_list): New.
23955 (add_var_loc_to_decl): Add view param. Test it against last.
23956 (new_loc_list): Add view params. Record them.
23957 (AT_loc_list_ptr): Handle loc and view lists.
23958 (view_list_to_loc_list_val_node): New.
23959 (print_dw_val): Handle dw_val_class_view_list.
23960 (size_of_die): Likewise.
23961 (value_format): Likewise.
23962 (loc_list_has_views): New.
23963 (gen_llsym): Set vl_symbol too.
23964 (maybe_gen_llsym, skip_loc_list_entry): New.
23965 (dwarf2out_maybe_output_loclist_view_pair): New.
23966 (output_loc_list): Output view list or entries too.
23967 (output_view_list_offset): New.
23968 (output_die): Handle dw_val_class_view_list.
23969 (output_dwarf_version): New.
23970 (output_compilation_unit_header): Use it.
23971 (output_skeleton_debug_sections): Likewise.
23972 (output_rnglists, output_line_info): Likewise.
23973 (output_pubnames, output_aranges): Update version comments.
23974 (output_one_line_info_table): Output view numbers in asm comments.
23975 (dw_loc_list): Determine current endview, pass it to new_loc_list.
23976 Call maybe_gen_llsym.
23977 (loc_list_from_tree_1): Adjust.
23978 (add_AT_location_description): Create view list attribute if
23979 needed, check it's absent otherwise.
23980 (convert_cfa_to_fb_loc_list): Adjust.
23981 (maybe_emit_file): Call output_asm_line_debug_info for test.
23982 (dwarf2out_var_location): Reset views as needed. Precompute
23983 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
23984 attribute. Set view.
23985 (new_line_info_table): Reset next view.
23986 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
23987 (dwarf2out_source_line): Likewise. Output view resets and labels to
23988 the assembler, or select appropriate line info opcodes.
23989 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
23990 (optimize_string_length): Catch it. Adjust.
23991 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
23992 dw_val_class_view_list, and remove it if no longer needed.
23993 (hash_loc_list): Hash view numbers.
23994 (loc_list_hasher::equal): Compare them.
23995 (optimize_location_lists): Check whether a view list symbol is
23996 needed, and whether the locview attribute is present, and
23997 whether they match. Remove the locview attribute if no longer
23999 (index_location_lists): Call skip_loc_list_entry for test.
24000 (dwarf2out_finish): Call output_asm_line_debug_info for test.
24001 Use output_dwarf_version.
24002 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
24003 (struct dw_val_node): Add val_view_list.
24004 * final.c (SEEN_NEXT_VIEW): New.
24005 (set_next_view_needed): New.
24006 (clear_next_view_needed): New.
24007 (maybe_output_next_view): New.
24008 (final_start_function): Rename to...
24009 (final_start_function_1): ... this. Take pointer to FIRST,
24010 add SEEN parameter. Emit param bindings in the initial view.
24011 (final_start_function): Reintroduce SEEN-less interface.
24012 (final): Rename to...
24013 (final_1): ... this. Take SEEN parameter. Output final pending
24014 next view at the end.
24015 (final): Reintroduce seen-less interface.
24016 (final_scan_insn): Output pending next view before switching
24017 sections or ending a block. Mark the next view as needed when
24018 outputting variable locations. Notify debug backend of section
24019 changes, and of location view changes.
24020 (rest_of_handle_final): Adjust.
24021 * toplev.c (process_options): Autodetect value for debug variable
24022 location views option. Warn on incompat5 without -gdwarf-5.
24023 * doc/invoke.texi (gvariable-location-views): New.
24024 (gvariable-location-views=incompat5): New.
24025 (gno-variable-location-views): New.
24027 2018-02-08 David Malcolm <dmalcolm@redhat.com>
24029 PR tree-optimization/84136
24030 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
24031 that the result of find_edge is non-NULL.
24033 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
24036 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
24037 storing integer register in SImode. Fix cost of 256 and 512
24038 byte aligned SSE register store.
24040 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
24042 * config/i386/i386.c (ix86_multiplication_cost): Fix
24043 multiplication cost for TARGET_AVX512DQ.
24045 2018-02-08 Marek Polacek <polacek@redhat.com>
24047 PR tree-optimization/84238
24048 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
24051 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
24053 PR tree-optimization/84265
24054 * tree-vect-stmts.c (vectorizable_store): Don't treat
24055 VMAT_CONTIGUOUS accesses as grouped.
24056 (vectorizable_load): Likewise.
24058 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
24060 PR tree-optimization/81635
24061 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
24062 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
24063 (test_round_for_mask): New functions.
24064 (wide_int_cc_tests): Call test_round_for_mask.
24065 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
24066 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
24067 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
24068 range returned by get_range_info.
24070 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
24073 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
24074 * symtab.c: Include builtins.h
24075 (symtab_node::output_to_lto_symbol_table_p): Move here
24076 from lto-streamer-out.c:output_symbol_p.
24077 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
24078 (output_symbol_p): Move all logic to symtab.c
24079 (produce_symtab): Update.
24081 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24083 * config/s390/s390-opts.h (enum indirect_branch): Define.
24084 * config/s390/s390-protos.h (s390_return_addr_from_memory)
24085 (s390_indirect_branch_via_thunk)
24086 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
24087 (enum s390_indirect_branch_type): Define.
24088 * config/s390/s390.c (struct s390_frame_layout, struct
24089 machine_function): Remove.
24090 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
24091 (indirect_branch_table_label_no, indirect_branch_table_name):
24093 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
24094 (enum s390_indirect_branch_option): Define.
24095 (s390_return_addr_from_memory): New function.
24096 (s390_handle_string_attribute): New function.
24097 (s390_attribute_table): Add new attribute handler.
24098 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
24099 (s390_indirect_branch_via_thunk): New function.
24100 (s390_indirect_branch_via_inline_thunk): New function.
24101 (s390_function_ok_for_sibcall): When jumping via thunk disallow
24102 sibling call optimization for non z10 compiles.
24103 (s390_emit_call): Force indirect branch target to be a single
24104 register. Add r1 clobber for non-z10 compiles.
24105 (s390_emit_epilogue): Emit return jump via return_use expander.
24106 (s390_reorg): Handle JUMP_INSNs as execute targets.
24107 (s390_option_override_internal): Perform validity checks for the
24108 new command line options.
24109 (s390_indirect_branch_attrvalue): New function.
24110 (s390_indirect_branch_settings): New function.
24111 (s390_set_current_function): Invoke s390_indirect_branch_settings.
24112 (s390_output_indirect_thunk_function): New function.
24113 (s390_code_end): Implement target hook.
24114 (s390_case_values_threshold): Implement target hook.
24115 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
24117 * config/s390/s390.h (struct s390_frame_layout)
24118 (struct machine_function): Move here from s390.c.
24119 (TARGET_INDIRECT_BRANCH_NOBP_RET)
24120 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
24121 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
24122 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
24123 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
24124 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
24125 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
24126 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
24127 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
24128 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
24129 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
24130 (mnemonic attribute): Add values which aren't recognized
24132 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
24133 pattern for branch conversion. Fix mnemonic attribute.
24134 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
24135 indirect branch via thunk if requested.
24136 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
24137 ("*indirect_jump"): Disable for branch conversion using out of
24139 ("indirect_jump_via_thunk<mode>_z10")
24140 ("indirect_jump_via_thunk<mode>")
24141 ("indirect_jump_via_inlinethunk<mode>_z10")
24142 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
24143 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
24144 ("casesi_jump_via_inlinethunk<mode>_z10")
24145 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
24146 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
24147 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
24148 ("*indirect2_jump"): Disable for branch conversion.
24149 ("casesi_jump"): Turn into expander and expand patterns for branch
24151 ("return_use"): New expander.
24152 ("*return"): Emit return via thunk and rename it to ...
24153 ("*return<mode>"): ... this one.
24154 * config/s390/s390.opt: Add new options and and enum for the
24157 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
24159 * lra-constraints.c (match_reload): Unconditionally use
24160 gen_lowpart_SUBREG, rather than selecting between that
24161 and equivalent gen_rtx_SUBREG code.
24163 2018-02-08 Richard Biener <rguenther@suse.de>
24165 PR tree-optimization/84233
24166 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
24167 changed flag instead of boguously re-using phi_inserted.
24169 2018-02-08 Martin Jambor <mjambor@suse.cz>
24171 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
24172 static local variables.
24174 2018-02-08 Richard Biener <rguenther@suse.de>
24176 PR tree-optimization/84278
24177 * tree-vect-stmts.c (vectorizable_store): When looking for
24178 smaller vector types to perform grouped strided loads/stores
24179 make sure the mode is supported by the target.
24180 (vectorizable_load): Likewise.
24182 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
24184 * config/aarch64/aarch64.c (aarch64_components_for_bb):
24185 Increase LDP/STP opportunities by adding adjacent callee-saves.
24187 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
24189 PR rtl-optimization/84068
24190 PR rtl-optimization/83459
24191 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
24193 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
24195 PR tree-optimization/84224
24196 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
24197 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
24198 non-zero arguments.
24200 2018-02-07 Iain Sandoe <iain@codesourcery.com>
24203 * config/rs6000/altivec.md (*restore_world): Remove LR use.
24204 * config/rs6000/predicates.md (restore_world_operation): Adjust op
24205 count, remove one USE.
24207 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
24209 * doc/install.texi (Configuration): Document the
24210 --with-long-double-format={ibm,ieee} PowerPC configuration
24214 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
24215 Convert from define_expand to be define_insn_and_split. Rework
24216 float/double/_Float128 conversions to QI/HI/SImode to work with
24217 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
24218 conversions to QI/HImode types did a store and then a load to
24219 truncate the value. For conversions to VSX registers, don't split
24220 the insn, instead emit the code directly. Use the code iterator
24221 any_fix to combine signed and unsigned conversions.
24222 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
24223 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
24224 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
24225 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
24226 (fix_<mode>di2_hw): Likewise.
24227 (fixuns_<mode>di2_hw): Likewise.
24228 (fix_<mode>si2_hw): Likewise.
24229 (fixuns_<mode>si2_hw): Likewise.
24230 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
24231 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
24232 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
24233 fix<uns>_trunc<SFDF:mode>si2_p8.
24234 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
24235 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
24236 (fix<uns>_<mode>_mem): Likewise.
24237 (fctiw<u>z_<mode>_mem): Likewise.
24238 (fix<uns>_<mode>_mem): Likewise.
24239 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
24240 the register allocator from doing a direct move to the GPRs to do
24241 a store, and instead use the ISA 3.0 store byte/half-word from
24242 vector register instruction. For IEEE 128-bit floating point,
24243 also optimize stores of 32-bit ints.
24244 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
24246 2018-02-07 Alan Hayward <alan.hayward@arm.com>
24248 * genextract.c (push_pathstr_operand): New function to support
24250 (walk_rtx): Call push_pathstr_operand.
24251 (print_path): Support [a-zA-Z].
24253 2018-02-07 Richard Biener <rguenther@suse.de>
24255 PR tree-optimization/84037
24256 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
24257 (cse_and_gimplify_to_preheader): Declare.
24258 (vect_get_place_in_interleaving_chain): Likewise.
24259 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
24261 (_loop_vec_info::~_loop_vec_info): Delete it.
24262 (cse_and_gimplify_to_preheader): New function.
24263 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
24264 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
24265 (vectorizable_load): Likewise. For grouped stores always base
24266 the IV on the first element.
24267 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
24268 condition before gimplifying.
24270 2018-02-07 Jakub Jelinek <jakub@redhat.com>
24272 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
24273 *DIV_EXPR and *MOD_EXPR.
24275 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
24278 * config/i386/i386.c (ix86_option_override_internal): Mask out
24279 the CF_SET bit when checking -fcf-protection.
24281 2018-02-07 Tom de Vries <tom@codesourcery.com>
24284 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
24287 2018-02-07 Richard Biener <rguenther@suse.de>
24289 PR tree-optimization/84204
24290 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
24293 PR tree-optimization/84205
24294 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
24295 special-case isl_ast_op_zdiv_r.
24297 PR tree-optimization/84223
24298 * graphite-scop-detection.c (gather_bbs::before_dom_children):
24299 Only add conditions from within the region.
24300 (gather_bbs::after_dom_children): Adjust.
24302 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
24305 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
24306 * config/avr/avr.md: Only post-reload split REG-REG moves if
24307 either register is GENERAL_REG_P.
24309 2018-02-07 Jakub Jelinek <jakub@redhat.com>
24311 PR tree-optimization/84235
24312 * tree-ssa-scopedtables.c
24313 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
24314 if the subtraction is performed in floating point type where NaNs are
24315 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
24316 build 1. Formatting fix.
24318 2018-02-06 Jakub Jelinek <jakub@redhat.com>
24321 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
24322 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
24323 and skip it regardless of bb boundaries. Use CALL_P macro,
24324 don't test INSN_P (insn) together with CALL_P or JUMP_P check
24325 unnecessarily, formatting fix.
24327 2018-02-06 Michael Collison <michael.collison@arm.com>
24329 * config/arm/thumb2.md:
24330 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
24331 (*thumb_mov_notscc): Ditto.
24333 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
24336 * config/rs6000/rs6000.md (su code attribute): Use "u" for
24337 unsigned_fix, not "s".
24339 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24341 * configure.ac (gcc_fn_eh_frame_ro): New function.
24342 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
24343 correct .eh_frame permissions.
24344 * configure: Regenerate.
24346 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
24348 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
24349 irrelevant options.
24351 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24353 * config/rs6000/rs6000.c (rs6000_option_override_internal):
24354 Display warning message for -mno-speculate-indirect-jumps.
24356 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
24358 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
24360 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
24362 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
24364 PR tree-optimization/84225
24365 * tree-eh.c (find_trapping_overflow): Only call
24366 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
24368 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
24371 * config/i386/i386.c: Reimplement the check of possible options
24372 -mibt/-mshstk conbination. Change error messages.
24373 * doc/invoke.texi: Fix a typo: remove extra '='.
24375 2018-02-06 Marek Polacek <polacek@redhat.com>
24377 PR tree-optimization/84228
24378 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
24380 2018-02-06 Tamar Christina <tamar.christina@arm.com>
24383 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
24384 emitted arch directives.
24385 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
24386 __ARM_FEATURE_COPROC before changing architectures.
24388 2018-02-06 Richard Biener <rguenther@suse.de>
24390 * config/i386/i386.c (print_reg): Fix typo.
24391 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
24393 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
24395 * configure: Regenerate.
24397 2018-02-05 Martin Sebor <msebor@redhat.com>
24399 PR tree-optimization/83369
24400 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
24403 2018-02-05 Martin Liska <mliska@suse.cz>
24405 * doc/invoke.texi: Cherry-pick upstream r323995.
24407 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
24409 * ira.c (ira_init_register_move_cost): Adjust comment.
24411 2018-02-05 Martin Liska <mliska@suse.cz>
24413 PR gcov-profile/84137
24414 * doc/gcov.texi: Fix typo in documentation.
24416 2018-02-05 Martin Liska <mliska@suse.cz>
24418 PR gcov-profile/83879
24419 * doc/gcov.texi: Document necessity of --dynamic-list-data when
24420 using dlopen functionality.
24422 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
24424 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
24425 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
24426 _mm_maskz_range_ss, _mm_mask_range_round_ss,
24427 _mm_maskz_range_round_ss): New intrinsics.
24428 (__builtin_ia32_rangesd128_round)
24429 (__builtin_ia32_rangess128_round): Remove.
24430 (__builtin_ia32_rangesd128_mask_round,
24431 __builtin_ia32_rangess128_mask_round): New builtins.
24432 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
24433 __builtin_ia32_rangess128_round): Remove.
24434 (__builtin_ia32_rangesd128_mask_round,
24435 __builtin_ia32_rangess128_mask_round): New builtins.
24436 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
24437 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
24438 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
24439 "<round_saeonly_constraint>")): Changed to ...
24440 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
24441 "<round_saeonly_scalar_constraint>")): ... this.
24442 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24443 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
24444 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
24445 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
24446 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
24448 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
24450 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
24452 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
24453 Remove all values except native, 8540 and 8548.
24455 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
24457 * config/i386/i386.c (ix86_output_function_return): Pass
24458 INVALID_REGNUM, instead of -1, as invalid register number to
24459 indirect_thunk_name and output_indirect_thunk.
24461 2018-02-02 Julia Koval <julia.koval@intel.com>
24463 * config.gcc: Add -march=icelake.
24464 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
24465 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
24466 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
24467 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
24468 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
24469 (processor_target_table): Add icelake.
24470 (ix86_option_override_internal): Handle new PTAs.
24471 (get_builtin_code_for_version): Handle icelake.
24472 (M_INTEL_COREI7_ICELAKE): New.
24473 (fold_builtin_cpu): Handle icelake.
24474 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
24475 * doc/invoke.texi: Add -march=icelake.
24477 2018-02-02 Julia Koval <julia.koval@intel.com>
24479 * config/i386/i386.c (ix86_option_override_internal): Change flags type
24480 to wide_int_bitmask.
24481 * wide-int-bitmask.h: New.
24483 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
24486 * config/i386/i386.md: Replace Pmode with word_mode in
24487 builtin_setjmp_setup and builtin_longjmp to support x32.
24489 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
24493 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
24495 (rs6000_supported_cpu_names): New static variable.
24496 (linux_cpu_translation_table): Likewise.
24497 (elf_platform) <cpu>: Define new static variable and use it.
24498 Translate kernel AT_PLATFORM name to canonical name if needed.
24499 Error if platform name is unknown.
24501 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
24504 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
24506 2018-02-01 Jeff Law <law@redhat.com>
24509 * config/i386/i386.c (release_scratch_register_on_entry): Add new
24510 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
24511 the scratch if RELEASE_VIA_POP is false.
24512 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
24513 If we have to save a temporary register, decrement SIZE appropriately.
24514 Pass new arguments to release_scratch_register_on_entry.
24515 (ix86_adjust_stack_and_probe): Likewise.
24516 (ix86_emit_probe_stack_range): Pass new arguments to
24517 release_scratch_register_on_entry.
24519 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
24521 PR rtl-optimization/84157
24522 * combine.c (change_zero_ext): Use REG_P predicate in
24523 front of HARD_REGISTER_P predicate.
24525 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
24527 * config/avr/avr.c (avr_option_override): Move disabling of
24528 -fdelete-null-pointer-checks to...
24529 * common/config/avr/avr-common.c (avr_option_optimization_table):
24532 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
24534 PR tree-optimization/81635
24535 * tree-data-ref.c (split_constant_offset_1): For types that
24536 wrap on overflow, try to use range info to prove that wrapping
24539 2018-02-01 Renlin Li <renlin.li@arm.com>
24542 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
24543 TAILCALL_ADDR_REGS.
24544 (aarch64_register_move_cost): Likewise.
24545 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
24546 TAILCALL_ADDR_REGS.
24547 (REG_CLASS_NAMES): Likewise.
24548 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
24549 TAILCALL_ADDR_REGS. Remove IP registers.
24550 * config/aarch64/aarch64.md (Ucs): Update register constraint.
24552 2018-02-01 Richard Biener <rguenther@suse.de>
24554 * domwalk.h (dom_walker::dom_walker): Add additional constructor
24555 for specifying RPO order and allow NULL for that.
24556 * domwalk.c (dom_walker::dom_walker): Likewise.
24557 (dom_walker::walk): Handle NULL RPO order.
24558 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
24560 (rewrite_update_dom_walker): Likewise.
24561 (mark_def_dom_walker): Likewise.
24563 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
24565 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
24566 (aarch64_maybe_expand_sve_subreg_move): Declare.
24567 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
24568 * config/aarch64/predicates.md (aarch64_any_register_operand): New
24570 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
24571 that are semantically a reverse operation.
24572 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
24573 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
24574 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
24576 (aarch64_can_change_mode_class): For big-endian, forbid changes
24577 between two SVE modes if they have different element sizes.
24579 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
24581 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
24582 the TImode handling for big-endian targets.
24584 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
24586 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
24587 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
24589 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
24590 Remove BSWAP handing for big-endian targets and use the form of
24591 LD1RQ appropariate for the mode.
24593 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
24595 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
24596 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
24597 duplicated element.
24599 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
24602 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
24603 check for operands that need to go through aarch64_sve_reload_be.
24605 2018-02-01 Jakub Jelinek <jakub@redhat.com>
24607 PR tree-optimization/81661
24608 PR tree-optimization/84117
24609 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
24610 * tree-eh.c: Include gimplify.h.
24611 (find_trapping_overflow, replace_trapping_overflow,
24612 rewrite_to_non_trapping_overflow): New functions.
24613 * tree-vect-loop.c: Include tree-eh.h.
24614 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
24615 * tree-data-ref.c: Include tree-eh.h.
24616 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
24618 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
24620 PR rtl-optimization/84123
24621 * combine.c (change_zero_ext): Check if hard register satisfies
24622 can_change_dest_mode before calling gen_lowpart_SUBREG.
24624 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
24627 * ira.c (ira_init_register_move_cost): Remove assert.
24629 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
24631 PR rtl-optimization/84071
24632 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
24633 * doc/tm.texi: Regenerate.
24635 2018-01-31 Richard Biener <rguenther@suse.de>
24637 PR tree-optimization/84132
24638 * tree-data-ref.c (analyze_miv_subscript): Properly
24639 check whether evolution_function_is_affine_multivariate_p
24640 before calling gcd_of_steps_may_divide_p.
24642 2018-01-31 Julia Koval <julia.koval@intel.com>
24645 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
24646 * config/i386/i386.md (rdpid_rex64) New.
24647 (rdpid): Make 32bit only.
24649 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
24652 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
24653 an IDENTIFIER_NODE for FUNCTION_TYPE's.
24655 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
24658 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
24660 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
24662 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
24664 PR rtl-optimization/84071
24665 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
24666 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
24668 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
24670 * config/arc/arc.c (arc_handle_aux_attribute): New function.
24671 (arc_attribute_table): Add 'aux' attribute.
24672 (arc_in_small_data_p): Consider aux like variables.
24673 (arc_is_aux_reg_p): New function.
24674 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
24675 (arc_get_aux_arg): New function.
24676 (prepare_move_operands): Handle aux-register access.
24677 (arc_handle_aux_attribute): New function.
24678 * doc/extend.texi (ARC Variable attributes): Add subsection.
24680 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
24682 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
24683 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
24684 (arc_attribute_table): Add 'uncached' attribute.
24685 (arc_print_operand): Print '.di' flag for uncached memory
24687 (arc_in_small_data_p): Do not consider for small data the uncached
24689 (arc_is_uncached_mem_p): New function.
24690 * config/arc/predicates.md (compact_store_memory_operand): Check
24691 for uncached memory accesses.
24692 (nonvol_nonimm_operand): Likewise.
24693 * doc/extend.texi (ARC Type Attribute): New subsection.
24695 2018-01-31 Jakub Jelinek <jakub@redhat.com>
24698 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
24699 falign-loops=): Add Optimization flag.
24701 2018-01-30 Jeff Law <law@redhat.com>
24704 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
24705 INT_REGISTERS_SAVED. Check it prior to calling
24706 get_scratch_register_on_entry.
24707 (ix86_adjust_stack_and_probe): Similarly.
24708 (ix86_emit_probe_stack_range): Similarly.
24709 (ix86_expand_prologue): Corresponding changes.
24711 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24714 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
24715 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
24717 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
24720 * lra-constraints.c (curr_insn_transform): Process AND in the
24723 2018-01-30 Jakub Jelinek <jakub@redhat.com>
24725 PR rtl-optimization/83986
24726 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
24727 dependence against last_pending_memory_flush in addition to
24728 pending_jump_insns.
24730 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
24732 PR tree-optimization/81611
24733 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
24736 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24739 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
24742 2018-01-30 Richard Biener <rguenther@suse.de>
24743 Jakub Jelinek <jakub@redhat.com>
24745 PR tree-optimization/84111
24746 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
24747 inner loops added during recursion, as they don't have up-to-date
24750 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
24753 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
24754 (can_inline_edge_by_limits_p): ... here.
24755 (can_early_inline_edge_p, check_callers,
24756 update_caller_keys, update_callee_keys, recursive_inlining,
24757 add_new_edges_to_heap, speculation_useful_p,
24758 inline_small_functions,
24759 inline_small_functions, flatten_function,
24760 inline_to_all_callers_1): Update.
24762 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
24764 * profile-count.c (profile_count::combine_with_ipa_count): Handle
24767 2018-01-30 Richard Biener <rguenther@suse.de>
24769 PR tree-optimization/83008
24770 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
24771 invariant and constant vector uses in stmts when they need
24772 more than one stmt.
24774 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24777 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
24778 * configure: Regenerate.
24780 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
24782 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
24784 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
24785 Use gen_rtx_REG rather than gen_lowpart.
24787 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
24789 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
24790 rather than 0 when creating partial subregs.
24792 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
24794 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
24797 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
24800 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
24801 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
24802 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
24803 flags. This restores the settings used before the 2017-07-24.
24804 Turning off pre increment/decrement/modify allows IVOPTS to
24805 optimize DF/SF loops where the index is an int.
24807 2018-01-29 Richard Biener <rguenther@suse.de>
24808 Kelvin Nilsen <kelvin@gcc.gnu.org>
24811 * tree-vect-stmts.c (vectorizable_call): Don't call
24812 targetm.vectorize_builtin_md_vectorized_function if callee is
24815 2018-01-22 Carl Love <cel@us.ibm.com>
24817 * doc/extend.tex: Fix typo in second arg in
24818 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
24820 2018-01-29 Richard Biener <rguenther@suse.de>
24822 PR tree-optimization/84086
24823 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
24824 (flush_ssaname_freelist): When SSA names were released reset
24825 the SCEV hash table.
24827 2018-01-29 Richard Biener <rguenther@suse.de>
24829 PR tree-optimization/84057
24830 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
24831 removed paths when removing edges.
24833 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
24835 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
24836 -mfunction-return=@var{choice}.
24838 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
24840 PR diagnostic/84034
24841 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
24842 Handle CR like TAB.
24843 (layout::print_source_line): Likewise.
24844 (test_get_line_width_without_trailing_whitespace): Add test cases.
24846 2018-01-27 Jakub Jelinek <jakub@redhat.com>
24848 PR middle-end/84040
24849 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
24852 2018-01-26 Jim Wilson <jimw@sifive.com>
24854 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
24856 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
24859 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24861 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
24862 and CMP + SUB-immediate -> SUBS.
24864 2018-01-26 Martin Sebor <msebor@redhat.com>
24866 PR tree-optimization/83896
24867 * tree-ssa-strlen.c (get_string_len): Rename...
24868 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
24869 Avoid assuming length is constant.
24870 (handle_char_store): Use HOST_WIDE_INT for string length.
24872 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
24875 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
24876 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
24878 2018-01-26 Richard Biener <rguenther@suse.de>
24880 PR rtl-optimization/84003
24881 * dse.c (record_store): Only record redundant stores when
24882 the earlier store aliases at least all accesses the later one does.
24884 2018-01-26 Jakub Jelinek <jakub@redhat.com>
24886 PR rtl-optimization/83985
24887 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
24888 REG_CFA_RESTORE insns.
24889 (delete_unmarked_insns): Don't ignore separate shrink wrapping
24890 REG_CFA_RESTORE insns here.
24893 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
24894 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
24896 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24898 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
24899 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
24900 (arc_init): Likewise.
24901 (arc_override_options): Likewise.
24902 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
24904 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
24906 * config/arc/arc.h (TARGET_DBNZ): Define.
24907 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
24908 properly set the tune attribute.
24909 (dbnz): Use TARGET_DBNZ guard.
24910 * config/arc/arc.opt (mtune): Add core3 option.
24912 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24914 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
24915 recognize new pic like addresses.
24916 (arc_delegitimize_address): Clean up.
24918 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24920 * config/arc/arc-arches.def: Option mrf16 valid for all
24922 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
24923 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
24924 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
24925 * config/arc/arc-tables.opt: Regenerate.
24926 * config/arc/arc.c (arc_conditional_register_usage): Handle
24927 reduced register file case.
24928 (arc_file_start): Set must have build attributes.
24929 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
24930 mrf16 option value.
24931 * config/arc/arc.opt (mrf16): Add new option.
24932 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
24933 * config/arc/genmultilib.awk: Handle new mrf16 option.
24934 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
24935 * config/arc/t-multilib: Regenerate.
24936 * doc/invoke.texi (ARC Options): Document mrf16 option.
24938 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24940 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
24941 * config/arc/arc.c (arc_handle_secure_attribute): New function.
24942 (arc_attribute_table): Add 'secure_call' attribute.
24943 (arc_print_operand): Print secure call operand.
24944 (arc_function_ok_for_sibcall): Don't optimize tail calls when
24946 (arc_is_secure_call_p): New function. * config/arc/arc.md
24947 (call_i): Add support for sjli instruction.
24948 (call_value_i): Likewise.
24949 * config/arc/constraints.md (Csc): New constraint.
24951 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24952 John Eric Martin <John.Martin@emmicro-us.com>
24954 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
24955 * config/arc/arc.c (_arc_jli_section): New struct.
24956 (arc_jli_section): New type.
24957 (rc_jli_sections): New static variable.
24958 (arc_handle_jli_attribute): New function.
24959 (arc_attribute_table): Add jli_always and jli_fixed attribute.
24960 (arc_file_end): New function.
24961 (TARGET_ASM_FILE_END): Define.
24962 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
24963 (arc_add_jli_section): New function.
24964 (jli_call_scan): Likewise.
24965 (arc_reorg): Call jli_call_scan.
24966 (arc_output_addsi): Remove 'S' from printing asm operand.
24967 (arc_is_jli_call_p): New function.
24968 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
24970 (movhi_insn): Likewise.
24971 (movsi_insn): Likewise.
24972 (movsi_set_cc_insn): Likewise.
24973 (loadqi_update): Likewise.
24974 (load_zeroextendqisi_update): Likewise.
24975 (load_signextendqisi_update): Likewise.
24976 (loadhi_update): Likewise.
24977 (load_zeroextendhisi_update): Likewise.
24978 (load_signextendhisi_update): Likewise.
24979 (loadsi_update): Likewise.
24980 (loadsf_update): Likewise.
24981 (movsicc_insn): Likewise.
24982 (bset_insn): Likewise.
24983 (bxor_insn): Likewise.
24984 (bclr_insn): Likewise.
24985 (bmsk_insn): Likewise.
24986 (bicsi3_insn): Likewise.
24987 (cmpsi_cc_c_insn): Likewise.
24988 (movsi_ne): Likewise.
24989 (movsi_cond_exec): Likewise.
24990 (clrsbsi2): Likewise.
24991 (norm_f): Likewise.
25001 (call_i): Remove 'S' asm letter, add jli instruction.
25002 (call_value_i): Likewise.
25003 * config/arc/arc.op (mjli-always): New option.
25004 * config/arc/constraints.md (Cji): New constraint.
25005 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
25007 (subsf3_fpx): Likewise.
25008 (mulsf3_fpx): Likewise.
25009 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
25011 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
25012 function attrbutes.
25013 * doc/invoke.texi (ARC): Document mjli-always option.
25015 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
25017 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
25018 avoid addition with 0 and use incw and decw where possible.
25020 2018-01-26 Richard Biener <rguenther@suse.de>
25022 PR tree-optimization/81082
25023 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
25024 association if it requires casting to unsigned.
25025 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
25026 from fold_plusminus_mult_expr to catch important cases late when
25027 range info is available.
25029 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25031 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
25032 * configure.ac (hidden_linkonce): New test.
25033 * configure: Regenerate.
25034 * config.in: Regenerate.
25036 2018-01-26 Julia Koval <julia.koval@intel.com>
25038 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
25039 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
25040 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
25041 _mm_mask_bitshuffle_epi64_mask): Fix type.
25042 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
25043 USI_FTYPE_V4DI_V4DI_USI): Remove.
25044 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
25045 __builtin_ia32_vpshufbitqmb256_mask,
25046 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
25047 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
25048 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
25050 2018-01-26 Alan Modra <amodra@gmail.com>
25053 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
25054 UNSPEC_VBPERMQ. Sort other unspecs.
25056 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
25058 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
25060 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
25062 PR middle-end/83055
25063 * predict.c (drop_profile): Do not push/pop cfun; update also
25065 (handle_missing_profiles): Fix logic looking for zero profiles.
25067 2018-01-25 Jakub Jelinek <jakub@redhat.com>
25069 PR middle-end/83977
25070 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
25071 on functions with #pragma omp declare simd or functions with simd
25073 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
25074 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
25075 Remove trailing \n from warning_at calls.
25077 2018-01-25 Tom de Vries <tom@codesourcery.com>
25080 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
25081 for neutered workers.
25083 2018-01-24 Joseph Myers <joseph@codesourcery.com>
25086 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
25087 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
25089 2018-01-24 Jeff Law <law@redhat.com>
25092 * i386.c (get_probe_interval): Move to earlier point.
25093 (ix86_compute_frame_layout): If -fstack-clash-protection and
25094 the frame is larger than the probe interval, then use pushes
25095 to save registers rather than reg->mem moves.
25096 (ix86_expand_prologue): Remove conditional for int_registers_saved
25099 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
25102 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
25103 min/max for never referenced object.
25105 2018-01-24 Jakub Jelinek <jakub@redhat.com>
25107 PR middle-end/83977
25108 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
25110 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
25111 attributes from DECL_ATTRIBUTES (decl) without affecting
25112 DECL_ATTRIBUTES (current_function_decl).
25113 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
25114 functions with non-NULL DECL_ABSTRACT_ORIGIN.
25116 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
25118 PR tree-optimization/83979
25119 * fold-const.c (fold_comparison): Use constant_boolean_node
25120 instead of boolean_{true,false}_node.
25122 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
25124 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
25127 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25129 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
25130 Simplify the clause that sets the length attribute.
25131 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
25132 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
25133 clause that sets the length attribute.
25134 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
25136 2018-01-24 Tom de Vries <tom@codesourcery.com>
25139 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
25140 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
25141 Add strict parameter.
25142 (prevent_branch_around_nothing): Insert dummy insn between branch to
25143 label and label with no ptx insn inbetween.
25144 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
25146 2018-01-24 Tom de Vries <tom@codesourcery.com>
25149 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
25150 for neutered threads in warp.
25151 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
25153 2018-01-24 Richard Biener <rguenther@suse.de>
25155 PR tree-optimization/83176
25156 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
25159 2018-01-24 Richard Biener <rguenther@suse.de>
25161 PR tree-optimization/82819
25162 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
25163 code generating pluses that are no-ops in the target precision.
25165 2018-01-24 Richard Biener <rguenther@suse.de>
25167 PR middle-end/84000
25168 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
25170 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
25172 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
25173 to merge probabilities.
25174 * predict.c (probably_never_executed): Also mark as cold functions
25175 with global 0 profile and guessed local profile.
25176 * profile-count.c (profile_probability::combine_with_count): New
25178 * profile-count.h (profile_probability::operator*,
25179 profile_probability::operator*=, profile_probability::operator/,
25180 profile_probability::operator/=): Reduce precision to adjusted
25181 and set value to guessed on contradictory divisions.
25182 (profile_probability::combine_with_freq): Remove.
25183 (profile_probability::combine_wiht_count): Declare.
25184 (profile_count::force_nonzero):: Set to adjusted.
25185 (profile_count::probability_in):: Set quality to adjusted.
25186 * tree-ssa-tail-merge.c (replace_block_by): Use
25187 combine_with_count.
25189 2018-01-23 Andrew Waterman <andrew@sifive.com>
25190 Jim Wilson <jimw@sifive.com>
25192 * config/riscv/riscv.c (riscv_stack_boundary): New.
25193 (riscv_option_override): Set riscv_stack_boundary. Handle
25194 riscv_preferred_stack_boundary_arg.
25195 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
25196 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
25197 (STACK_BOUNDARY): Set to riscv_stack_boundary.
25198 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
25199 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
25200 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
25202 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
25205 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
25206 of struct ix86_frame.
25207 (ix86_expand_epilogue): Likewise. Add a local variable for
25208 the reg_save_offset field in struct ix86_frame.
25210 2018-01-23 Bin Cheng <bin.cheng@arm.com>
25212 PR tree-optimization/82604
25213 * tree-loop-distribution.c (enum partition_kind): New enum item
25214 PKIND_PARTIAL_MEMSET.
25215 (partition_builtin_p): Support above new enum item.
25216 (generate_code_for_partition): Ditto.
25217 (compute_access_range): Differentiate cases that equality can be
25218 proven at all loops, the innermost loops or no loops.
25219 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
25220 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
25221 (finalize_partitions, distribute_loop): Don't fuse partition of
25222 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
25223 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
25224 parloop is enabled.
25226 2018-01-23 Martin Liska <mliska@suse.cz>
25228 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
25229 order to ignore the predictor.
25230 (PRED_POLYMORPHIC_CALL): Likewise.
25231 (PRED_RECURSIVE_CALL): Likewise.
25233 2018-01-23 Martin Liska <mliska@suse.cz>
25235 * tree-profile.c (tree_profiling): Print function header to
25236 aware reader which function we are working on.
25237 * value-prof.c (gimple_find_values_to_profile): Do not print
25238 not interesting value histograms.
25240 2018-01-23 Martin Liska <mliska@suse.cz>
25242 * profile-count.h (enum profile_quality): Add
25243 profile_uninitialized as the first value. Do not number values
25244 as they are zero based.
25245 (profile_count::verify): Update sanity check.
25246 (profile_probability::verify): Likewise.
25248 2018-01-23 Nathan Sidwell <nathan@acm.org>
25250 * doc/invoke.texi (ffor-scope): Deprecate.
25252 2018-01-23 David Malcolm <dmalcolm@redhat.com>
25254 PR tree-optimization/83510
25255 * domwalk.c (set_all_edges_as_executable): New function.
25256 (dom_walker::dom_walker): Convert bool param
25257 "skip_unreachable_blocks" to enum reachability. Move setup of
25258 edge flags to set_all_edges_as_executable and only do it when
25259 reachability is REACHABLE_BLOCKS.
25260 * domwalk.h (enum dom_walker::reachability): New enum.
25261 (dom_walker::dom_walker): Convert bool param
25262 "skip_unreachable_blocks" to enum reachability.
25263 (set_all_edges_as_executable): New decl.
25264 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
25265 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
25267 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
25268 but converting true to REACHABLE_BLOCKS.
25269 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
25271 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
25272 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
25273 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
25275 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
25276 if check_all_array_refs will be called.
25278 2018-01-23 David Malcolm <dmalcolm@redhat.com>
25280 * tree.c (selftest::test_location_wrappers): Add more test
25283 2018-01-23 David Malcolm <dmalcolm@redhat.com>
25285 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
25286 (selftest::test_bit_in_range): Likewise.
25288 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
25291 * doc/sourcebuild.texi (vect_float): Say that the selector
25292 only describes the situation when -funsafe-math-optimizations is on.
25293 (vect_float_strict): Document.
25295 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
25297 PR tree-optimization/83965
25298 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
25299 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
25300 instead of checking only for a reduction.
25301 (vect_recog_widen_sum_pattern): Likewise.
25303 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
25305 * predict.c (probably_never_executed): Only use precise profile info.
25306 (compute_function_frequency): Skip after inlining hack since we now
25307 have quality checking.
25309 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
25311 * profile-count.h (profile_probability::very_unlikely,
25312 profile_probability::unlikely, profile_probability::even): Set
25313 precision to guessed.
25315 2018-01-23 Richard Biener <rguenther@suse.de>
25317 PR tree-optimization/83963
25318 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
25319 Properly terminate dominator walk when crossing the exit edge not
25320 when visiting its source block.
25322 2018-01-23 Jakub Jelinek <jakub@redhat.com>
25325 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
25326 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
25328 2018-01-22 Jakub Jelinek <jakub@redhat.com>
25330 PR tree-optimization/83957
25331 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
25332 semicolon after for body surrounded by braces.
25334 PR tree-optimization/83081
25335 * profile-count.h (profile_probability::split): New method.
25336 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
25337 Use profile_probability::split.
25338 (do_compare_rtx_and_jump): Fix adjustment of probabilities
25339 when splitting a single conditional jump into 2.
25341 2018-01-22 David Malcolm <dmalcolm@redhat.com>
25343 PR tree-optimization/69452
25344 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
25347 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25349 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
25350 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
25351 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
25353 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25355 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
25357 * config/rl78/rl78.md (movdi): New define_expand.
25358 * config/rl78/rl78.c (rl78_split_movdi): New function.
25360 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
25363 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
25365 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
25366 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
25367 128-bit to produce an UNSPEC move to get the double word with the
25368 signbit and then a shift directly to do signbit.
25369 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
25370 implementation with a new version that just does either a direct
25371 move or a regular move. Move memory interface to separate insns.
25372 Move insns so they are next to the expander.
25373 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
25374 with signbit move. Split big and little endian case.
25375 (signbit<mode>2_dm_mem_le): Likewise.
25376 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
25377 (signbit<mode>2_dm2): Likewise.
25379 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25381 * config/rl78/rl78.md (anddi3): New define_expand.
25383 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25385 * config/rl78/rl78.md (umindi3): New define_expand.
25387 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25389 * config/rl78/rl78.md (smindi3): New define_expand.
25391 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25393 * config/rl78/rl78.md (smaxdi3): New define_expand.
25395 2018-01-22 Carl Love <cel@us.ibm.com>
25397 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
25398 LVX_V1TI): Add macro expansion.
25399 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
25400 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
25401 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
25402 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
25403 Change check to determine if the instruction is a byte reversing
25404 entry. Fix typo in comment.
25405 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
25406 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
25407 Add def_builtin calls for new builtins.
25408 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
25409 Add define_insn expansion.
25411 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25413 * config/rl78/rl78.md (umaxdi3): New define_expand.
25415 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25417 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
25418 for non-QImode registers.
25420 2018-01-22 Richard Biener <rguenther@suse.de>
25422 PR tree-optimization/83963
25423 * graphite-scop-detection.c (scop_detection::get_sese): Delay
25424 including the loop exit block.
25425 (scop_detection::merge_sese): Likewise.
25426 (scop_detection::add_scop): Do it here instead.
25428 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25430 * doc/sourcebuild.texi (arm_softfloat): Document.
25432 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
25435 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
25436 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
25437 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
25439 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25440 David Edelsohn <dje.gcc@gmail.com>
25443 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
25444 Change "crset eq" to "crset 2".
25445 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
25446 (*call_indirect_aix<mode>_nospec): Likewise.
25447 (*call_value_indirect_aix<mode>_nospec): Likewise.
25448 (*call_indirect_elfv2<mode>_nospec): Likewise.
25449 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
25450 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
25451 change assembly output from . to $.
25452 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
25453 (indirect_jump<mode>_nospec): Change assembly output from . to $.
25454 (*tablejump<mode>_internal1_nospec): Likewise.
25456 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
25459 * config/sh/sh_optimize_sett_clrt.cc:
25460 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
25462 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
25464 PR tree-optimization/83940
25465 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
25466 offset_dt to vect_constant_def rather than vect_unknown_def_type.
25467 (vect_check_load_store_mask): Add a mask_dt_out parameter and
25468 use it to pass back the definition type.
25469 (vect_check_store_rhs): Likewise rhs_dt_out.
25470 (vect_build_gather_load_calls): Add a mask_dt argument and use
25471 it instead of a call to vect_is_simple_use.
25472 (vectorizable_store): Update calls to vect_check_load_store_mask
25473 and vect_check_store_rhs. Use the dt returned by the latter instead
25474 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
25475 instead of calls to vect_is_simple_use. Pass the scalar rather
25476 than the vector operand to vect_is_simple_use when handling
25477 second and subsequent copies of an rhs value.
25478 (vectorizable_load): Update calls to vect_check_load_store_mask
25479 and vect_build_gather_load_calls. Use the cached mask_dt and
25480 gs_info.offset_dt instead of calls to vect_is_simple_use.
25482 2018-01-20 Jakub Jelinek <jakub@redhat.com>
25484 PR middle-end/83945
25485 * tree-emutls.c: Include gimplify.h.
25486 (lower_emutls_2): New function.
25487 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
25488 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
25489 it before further processing.
25492 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
25493 UINTVAL (trueop1) instead of INTVAL (op1).
25495 2018-01-19 Jakub Jelinek <jakub@redhat.com>
25499 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
25500 INCOMING_FRAME_SP_OFFSET if not defined.
25501 (scan_trace): Add ENTRY argument. If true and
25502 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
25503 emit a note to adjust the CFA offset.
25504 (create_cfi_notes): Adjust scan_trace callers.
25505 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
25506 INCOMING_FRAME_SP_OFFSET in the CIE.
25507 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
25508 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
25510 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
25511 * doc/tm.texi: Regenerated.
25513 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25515 PR rtl-optimization/83147
25516 * lra-constraints.c (remove_inheritance_pseudos): Use
25517 lra_substitute_pseudo_within_insn.
25519 2018-01-19 Tom de Vries <tom@codesourcery.com>
25520 Cesar Philippidis <cesar@codesourcery.com>
25523 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
25525 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
25528 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
25529 spaces for function labels.
25531 2018-01-19 Martin Liska <mliska@suse.cz>
25533 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
25534 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
25535 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
25536 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
25537 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
25538 (PRED_CONST_RETURN): Change from 69 to 65.
25539 (PRED_NULL_RETURN): Change from 91 to 71.
25540 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
25541 (PRED_LOOP_GUARD): Change from 66 to 73.
25543 2018-01-19 Martin Liska <mliska@suse.cz>
25545 * predict.c (predict_insn_def): Add new assert.
25546 (struct branch_predictor): Change type to signed integer.
25547 (test_prediction_value_range): Amend test to cover
25548 PROB_UNINITIALIZED.
25549 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
25550 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
25551 (PRED_LOOP_ITERATIONS_MAX): Likewise.
25552 (PRED_LOOP_IV_COMPARE): Likewise.
25553 * predict.h (PROB_UNINITIALIZED): Define new constant.
25555 2018-01-19 Martin Liska <mliska@suse.cz>
25557 * predict.c (dump_prediction): Add new format for
25558 analyze_brprob.py script which is enabled with -details
25560 * profile-count.h (precise_p): New function.
25562 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
25564 PR tree-optimization/83922
25565 * tree-vect-loop.c (vect_verify_full_masking): Return false if
25566 there are no statements that need masking.
25567 (vect_active_double_reduction_p): New function.
25568 (vect_analyze_loop_operations): Use it when handling phis that
25569 are not in the loop header.
25571 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
25573 PR tree-optimization/83914
25574 * tree-vect-loop.c (vectorizable_induction): Don't convert
25575 init_expr or apply the peeling adjustment for inductions
25576 that are nested within the vectorized loop.
25578 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25580 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
25583 2018-01-18 Jakub Jelinek <jakub@redhat.com>
25587 * function.h (gimplify_parameters): Add gimple_seq * argument.
25588 * function.c: Include gimple.h and options.h.
25589 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
25590 for the added local temporaries if needed.
25591 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
25592 if there are any parameter cleanups, wrap whole body into a
25593 try/finally with the cleanups.
25595 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
25598 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
25599 Use GET_MODE_CLASS for scalar floating point.
25601 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
25605 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
25606 Fix call of call_cgraph_insertion_hooks.
25608 2018-01-18 Martin Sebor <msebor@redhat.com>
25610 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
25612 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
25615 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
25618 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
25621 * common.opt: (-ffile-prefix-map): New option.
25622 * opts.c (common_handle_option): Defer it.
25623 * opts-global.c (handle_common_deferred_options): Handle it.
25624 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
25625 * file-prefix-map.h: New file.
25626 (remap_debug_filename, add_debug_prefix_map): ...here.
25627 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
25628 * final.c (debug_prefix_map, add_debug_prefix_map
25629 remap_debug_filename): Move to...
25630 * file-prefix-map.c: New file.
25631 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
25632 generalize, get rid of alloca(), use strrchr() instead of strchr().
25633 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
25634 Implement in terms of add_prefix_map().
25635 (remap_macro_filename, remap_debug_filename): Implement in term of
25637 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
25638 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
25639 * dbxout.c: Include file-prefix-map.h.
25640 * varasm.c: Likewise.
25641 * vmsdbgout.c: Likewise.
25642 * xcoffout.c: Likewise.
25643 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
25644 * doc/cppopts.texi (-fmacro-prefix-map): Document.
25645 * doc/invoke.texi (-ffile-prefix-map): Document.
25646 (-fdebug-prefix-map): Update description.
25648 2018-01-18 Martin Liska <mliska@suse.cz>
25650 * config/i386/i386.c (indirect_thunk_name): Document that also
25652 (output_indirect_thunk): Document why both instructions
25653 (pause and lfence) are generated.
25655 2018-01-18 Richard Biener <rguenther@suse.de>
25657 PR tree-optimization/83887
25658 * graphite-scop-detection.c
25659 (scop_detection::get_nearest_dom_with_single_entry): Remove.
25660 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
25661 (scop_detection::merge_sese): Re-implement with a flood-fill
25662 algorithm that properly finds a SESE region if it exists.
25664 2018-01-18 Jakub Jelinek <jakub@redhat.com>
25667 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
25668 pointer_diff optimizations use view_convert instead of convert.
25670 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25672 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
25673 Generate different code for -mno-speculate-indirect-jumps.
25674 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
25675 (*call_indirect_aix<mode>): Disable for
25676 -mno-speculate-indirect-jumps.
25677 (*call_indirect_aix<mode>_nospec): New define_insn.
25678 (*call_value_indirect_aix<mode>): Disable for
25679 -mno-speculate-indirect-jumps.
25680 (*call_value_indirect_aix<mode>_nospec): New define_insn.
25681 (*sibcall_nonlocal_sysv<mode>): Generate different code for
25682 -mno-speculate-indirect-jumps.
25683 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
25685 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
25687 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
25688 long double type, set the flags for noting the default long double
25689 type, even if we don't pass or return a long double type.
25691 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
25694 * ipa-inline.c (flatten_function): Do not overwrite final inlining
25697 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
25699 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
25700 support for merge[hl].
25701 (fold_mergehl_helper): New helper function.
25702 (tree-vector-builder.h): New #include for tree_vector_builder usage.
25703 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
25704 (altivec_vmrglw_direct): Add xxmrglw insn.
25706 2018-01-17 Andrew Waterman <andrew@sifive.com>
25708 * config/riscv/riscv.c (riscv_conditional_register_usage): If
25709 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
25711 2018-01-17 David Malcolm <dmalcolm@redhat.com>
25714 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
25715 call the lto_location_cache before reading the
25716 DECL_SOURCE_LOCATION of the types.
25718 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
25719 Richard Sandiford <richard.sandiford@linaro.org>
25721 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
25722 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
25723 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
25724 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
25725 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
25727 * config/aarch64/constraints.md (aarch64_movti_operand):
25729 * config/aarch64/predicates.md (Uti): Add new constraint.
25731 2018-01-17 Carl Love <cel@us.ibm.com>
25733 * config/rs6000/vsx.md (define_expand xl_len_r,
25734 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
25735 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
25737 (define_expand, define_insn): Move the shift left from the
25738 define_insn to the define_expand for lxvl and stxvl instructions.
25739 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
25740 and XL_LEN_R definitions to PURE.
25742 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
25744 * config/i386/i386.c (indirect_thunk_name): Declare regno
25745 as unsigned int. Compare regno with INVALID_REGNUM.
25746 (output_indirect_thunk): Ditto.
25747 (output_indirect_thunk_function): Ditto.
25748 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
25749 in the call to output_indirect_thunk_function.
25751 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
25753 PR middle-end/83884
25754 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
25755 rather than the size of inner_type to determine the stack slot size
25756 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
25758 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
25761 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
25764 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
25766 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
25767 endian Linux systems to optionally enable multilibs for selecting
25768 the long double type if the user configured an explicit type.
25769 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
25770 have no long double multilibs if not defined.
25771 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
25772 warn if the user used -mabi={ieee,ibm}longdouble and we built
25773 multilibs for long double.
25774 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
25775 appropriate multilib option.
25776 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
25778 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
25779 for building long double multilibs.
25780 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
25782 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
25784 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
25787 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
25789 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
25792 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
25795 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
25798 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
25800 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
25801 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
25803 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
25805 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
25806 different rtl trees depending on TARGET_64BIT.
25807 (rs6000_gen_lvx): Likewise.
25809 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
25811 * config/visium/visium.md (nop): Tweak comment.
25812 (hazard_nop): Likewise.
25814 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25816 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
25817 -mspeculate-indirect-jumps.
25818 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
25819 for -mno-speculate-indirect-jumps.
25820 (*call_indirect_elfv2<mode>_nospec): New define_insn.
25821 (*call_value_indirect_elfv2<mode>): Disable for
25822 -mno-speculate-indirect-jumps.
25823 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
25824 (indirect_jump): Emit different RTL for
25825 -mno-speculate-indirect-jumps.
25826 (*indirect_jump<mode>): Disable for
25827 -mno-speculate-indirect-jumps.
25828 (*indirect_jump<mode>_nospec): New define_insn.
25829 (tablejump): Emit different RTL for
25830 -mno-speculate-indirect-jumps.
25831 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
25832 (tablejumpsi_nospec): New define_expand.
25833 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
25834 (tablejumpdi_nospec): New define_expand.
25835 (*tablejump<mode>_internal1): Disable for
25836 -mno-speculate-indirect-jumps.
25837 (*tablejump<mode>_internal1_nospec): New define_insn.
25838 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
25841 2018-01-16 Artyom Skrobov tyomitch@gmail.com
25843 * caller-save.c (insert_save): Drop unnecessary parameter. All
25846 2018-01-16 Jakub Jelinek <jakub@redhat.com>
25847 Richard Biener <rguenth@suse.de>
25850 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
25851 return early, inline manually is_gimple_sizepos. Make sure if we
25852 call gimplify_expr we don't end up with a gimple constant.
25853 * tree.c (variably_modified_type_p): Don't return true for
25854 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
25855 * gimplify.h (is_gimple_sizepos): Remove.
25857 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25859 PR tree-optimization/83857
25860 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
25861 vectorizable_live_operation for pure SLP statements.
25862 (vectorizable_live_operation): Handle PHIs.
25864 2018-01-16 Richard Biener <rguenther@suse.de>
25866 PR tree-optimization/83867
25867 * tree-vect-stmts.c (vect_transform_stmt): Precompute
25868 nested_in_vect_loop_p since the scalar stmt may get invalidated.
25870 2018-01-16 Jakub Jelinek <jakub@redhat.com>
25873 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
25874 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
25875 If off is not INTEGER_CST, issue a may not be aligned warning
25876 rather than isn't aligned. Use isn%'t rather than isn't.
25877 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
25879 <case MULT_EXPR>: Improve the case when bottom and one of the
25880 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
25881 operand, in that case check if the other operand is multiple of
25882 bottom divided by the INTEGER_CST operand.
25884 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25887 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
25888 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
25889 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
25890 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
25891 * config/pa/pa.c (pa_function_arg_advance): Likewise.
25892 (pa_function_arg, pa_arg_partial_bytes): Likewise.
25893 (pa_function_arg_size): New function.
25895 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25897 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
25898 in a separate statement.
25900 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25902 PR tree-optimization/83847
25903 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
25904 group gathers and scatters.
25906 2018-01-16 Jakub Jelinek <jakub@redhat.com>
25908 PR rtl-optimization/86620
25909 * params.def (max-sched-ready-insns): Bump minimum value to 1.
25911 PR rtl-optimization/83213
25912 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
25913 to last if both are JUMP_INSNs.
25915 PR tree-optimization/83843
25916 * gimple-ssa-store-merging.c
25917 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
25918 store_immediate_info for bswap/nop orig_stores.
25920 2018-01-15 Andrew Waterman <andrew@sifive.com>
25922 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
25924 <UDIV>: Increase cost if !TARGET_DIV.
25926 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
25928 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
25929 (define_attr "cr_logical_3op"): New.
25930 (cceq_ior_compare): Adjust.
25931 (cceq_ior_compare_complement): Adjust.
25932 (*cceq_rev_compare): Adjust.
25933 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
25934 (is_cracked_insn): Adjust.
25935 (insn_must_be_first_in_group): Adjust.
25936 * config/rs6000/40x.md: Adjust.
25937 * config/rs6000/440.md: Adjust.
25938 * config/rs6000/476.md: Adjust.
25939 * config/rs6000/601.md: Adjust.
25940 * config/rs6000/603.md: Adjust.
25941 * config/rs6000/6xx.md: Adjust.
25942 * config/rs6000/7450.md: Adjust.
25943 * config/rs6000/7xx.md: Adjust.
25944 * config/rs6000/8540.md: Adjust.
25945 * config/rs6000/cell.md: Adjust.
25946 * config/rs6000/e300c2c3.md: Adjust.
25947 * config/rs6000/e500mc.md: Adjust.
25948 * config/rs6000/e500mc64.md: Adjust.
25949 * config/rs6000/e5500.md: Adjust.
25950 * config/rs6000/e6500.md: Adjust.
25951 * config/rs6000/mpc.md: Adjust.
25952 * config/rs6000/power4.md: Adjust.
25953 * config/rs6000/power5.md: Adjust.
25954 * config/rs6000/power6.md: Adjust.
25955 * config/rs6000/power7.md: Adjust.
25956 * config/rs6000/power8.md: Adjust.
25957 * config/rs6000/power9.md: Adjust.
25958 * config/rs6000/rs64.md: Adjust.
25959 * config/rs6000/titan.md: Adjust.
25961 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25963 * config/i386/predicates.md (indirect_branch_operand): Rewrite
25964 ix86_indirect_branch_register logic.
25966 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25968 * config/i386/constraints.md (Bs): Update
25969 ix86_indirect_branch_register check. Don't check
25970 ix86_indirect_branch_register with GOT_memory_operand.
25972 * config/i386/predicates.md (GOT_memory_operand): Don't check
25973 ix86_indirect_branch_register here.
25974 (GOT32_symbol_operand): Likewise.
25976 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25978 * config/i386/predicates.md (constant_call_address_operand):
25979 Rewrite ix86_indirect_branch_register logic.
25980 (sibcall_insn_operand): Likewise.
25982 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25984 * config/i386/constraints.md (Bs): Replace
25985 ix86_indirect_branch_thunk_register with
25986 ix86_indirect_branch_register.
25988 * config/i386/i386.md (indirect_jump): Likewise.
25989 (tablejump): Likewise.
25990 (*sibcall_memory): Likewise.
25991 (*sibcall_value_memory): Likewise.
25992 Peepholes of indirect call and jump via memory: Likewise.
25993 * config/i386/i386.opt: Likewise.
25994 * config/i386/predicates.md (indirect_branch_operand): Likewise.
25995 (GOT_memory_operand): Likewise.
25996 (call_insn_operand): Likewise.
25997 (sibcall_insn_operand): Likewise.
25998 (GOT32_symbol_operand): Likewise.
26000 2018-01-15 Jakub Jelinek <jakub@redhat.com>
26002 PR middle-end/83837
26003 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
26004 type rather than type addr's type points to.
26005 (expand_omp_atomic_mutex): Likewise.
26006 (expand_omp_atomic): Likewise.
26008 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
26011 * config/i386/i386.c (output_indirect_thunk_function): Use
26012 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
26013 for __x86_return_thunk.
26015 2018-01-15 Richard Biener <rguenther@suse.de>
26017 PR middle-end/83850
26018 * expmed.c (extract_bit_field_1): Fix typo.
26020 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26023 * config/arm/iterators.md (VF): New mode iterator.
26024 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
26025 Remove integer-related logic from pattern.
26026 (neon_vabd<mode>_3): Likewise.
26028 2018-01-15 Jakub Jelinek <jakub@redhat.com>
26030 PR middle-end/82694
26031 * common.opt (fstrict-overflow): No longer an alias.
26032 (fwrapv-pointer): New option.
26033 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
26034 also for pointer types based on flag_wrapv_pointer.
26035 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
26036 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
26037 opts->x_flag_wrapv got set.
26038 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
26039 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
26040 POINTER_TYPE_OVERFLOW_UNDEFINED.
26041 * match.pd: Likewise in address comparison pattern.
26042 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
26044 2018-01-15 Richard Biener <rguenther@suse.de>
26047 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
26048 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
26049 Reset type names to their identifier if their TYPE_DECL doesn't
26050 have linkage (and thus is used for ODR and devirt).
26051 (save_debug_info_for_decl): Remove.
26052 (save_debug_info_for_type): Likewise.
26053 (add_tree_to_fld_list): Adjust.
26054 * tree-pretty-print.c (dump_generic_node): Make dumping of
26055 type names more robust.
26057 2018-01-15 Richard Biener <rguenther@suse.de>
26059 * BASE-VER: Bump to 8.0.1.
26061 2018-01-14 Martin Sebor <msebor@redhat.com>
26064 * builtins.c (check_access): Avoid warning when the no-warning bit
26067 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
26069 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
26070 * ira-color (allocno_hard_regs_compare): Likewise.
26072 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
26075 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
26076 Use .pushsection/.popsection.
26078 2018-01-14 Martin Sebor <msebor@redhat.com>
26081 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
26083 2018-01-14 Jakub Jelinek <jakub@redhat.com>
26085 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
26086 entry from extra_headers.
26087 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
26088 extra_headers, make the list bitwise identical to the i?86-*-* one.
26090 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
26092 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
26093 -mcmodel=large with -mindirect-branch=thunk,
26094 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
26095 -mfunction-return=thunk-extern.
26096 * doc/invoke.texi: Document -mcmodel=large is incompatible with
26097 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
26098 -mfunction-return=thunk and -mfunction-return=thunk-extern.
26100 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
26102 * config/i386/i386.c (print_reg): Print the name of the full
26103 integer register without '%'.
26104 (ix86_print_operand): Handle 'V'.
26105 * doc/extend.texi: Document 'V' modifier.
26107 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
26109 * config/i386/constraints.md (Bs): Disallow memory operand for
26110 -mindirect-branch-register.
26112 * config/i386/predicates.md (indirect_branch_operand): Likewise.
26113 (GOT_memory_operand): Likewise.
26114 (call_insn_operand): Likewise.
26115 (sibcall_insn_operand): Likewise.
26116 (GOT32_symbol_operand): Likewise.
26117 * config/i386/i386.md (indirect_jump): Call convert_memory_address
26118 for -mindirect-branch-register.
26119 (tablejump): Likewise.
26120 (*sibcall_memory): Likewise.
26121 (*sibcall_value_memory): Likewise.
26122 Disallow peepholes of indirect call and jump via memory for
26123 -mindirect-branch-register.
26124 (*call_pop): Replace m with Bw.
26125 (*call_value_pop): Likewise.
26126 (*sibcall_pop_memory): Replace m with Bs.
26127 * config/i386/i386.opt (mindirect-branch-register): New option.
26128 * doc/invoke.texi: Document -mindirect-branch-register option.
26130 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
26132 * config/i386/i386-protos.h (ix86_output_function_return): New.
26133 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
26134 set function_return_type.
26135 (indirect_thunk_name): Add ret_p to indicate thunk for function
26137 (output_indirect_thunk_function): Pass false to
26138 indirect_thunk_name.
26139 (ix86_output_indirect_branch_via_reg): Likewise.
26140 (ix86_output_indirect_branch_via_push): Likewise.
26141 (output_indirect_thunk_function): Create alias for function
26142 return thunk if regno < 0.
26143 (ix86_output_function_return): New function.
26144 (ix86_handle_fndecl_attribute): Handle function_return.
26145 (ix86_attribute_table): Add function_return.
26146 * config/i386/i386.h (machine_function): Add
26147 function_return_type.
26148 * config/i386/i386.md (simple_return_internal): Use
26149 ix86_output_function_return.
26150 (simple_return_internal_long): Likewise.
26151 * config/i386/i386.opt (mfunction-return=): New option.
26152 (indirect_branch): Mention -mfunction-return=.
26153 * doc/extend.texi: Document function_return function attribute.
26154 * doc/invoke.texi: Document -mfunction-return= option.
26156 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
26158 * config/i386/i386-opts.h (indirect_branch): New.
26159 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
26160 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
26161 with local indirect jump when converting indirect call and jump.
26162 (ix86_set_indirect_branch_type): New.
26163 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
26164 (indirectlabelno): New.
26165 (indirect_thunk_needed): Likewise.
26166 (indirect_thunk_bnd_needed): Likewise.
26167 (indirect_thunks_used): Likewise.
26168 (indirect_thunks_bnd_used): Likewise.
26169 (INDIRECT_LABEL): Likewise.
26170 (indirect_thunk_name): Likewise.
26171 (output_indirect_thunk): Likewise.
26172 (output_indirect_thunk_function): Likewise.
26173 (ix86_output_indirect_branch_via_reg): Likewise.
26174 (ix86_output_indirect_branch_via_push): Likewise.
26175 (ix86_output_indirect_branch): Likewise.
26176 (ix86_output_indirect_jmp): Likewise.
26177 (ix86_code_end): Call output_indirect_thunk_function if needed.
26178 (ix86_output_call_insn): Call ix86_output_indirect_branch if
26180 (ix86_handle_fndecl_attribute): Handle indirect_branch.
26181 (ix86_attribute_table): Add indirect_branch.
26182 * config/i386/i386.h (machine_function): Add indirect_branch_type
26183 and has_local_indirect_jump.
26184 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
26186 (tablejump): Likewise.
26187 (*indirect_jump): Use ix86_output_indirect_jmp.
26188 (*tablejump_1): Likewise.
26189 (simple_return_indirect_internal): Likewise.
26190 * config/i386/i386.opt (mindirect-branch=): New option.
26191 (indirect_branch): New.
26194 (thunk-inline): Likewise.
26195 (thunk-extern): Likewise.
26196 * doc/extend.texi: Document indirect_branch function attribute.
26197 * doc/invoke.texi: Document -mindirect-branch= option.
26199 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
26202 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
26204 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
26206 * ipa-inline.c (want_inline_small_function_p): Return false if
26207 inlining has already failed with CIF_FINAL_ERROR.
26208 (update_caller_keys): Call want_inline_small_function_p before
26210 (update_callee_keys): Likewise.
26212 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26214 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
26216 (rs6000_quadword_masked_address_p): Likewise.
26217 (quad_aligned_load_p): Likewise.
26218 (quad_aligned_store_p): Likewise.
26219 (const_load_sequence_p): Add comment to describe the outer-most loop.
26220 (mimic_memory_attributes_and_flags): New function.
26221 (rs6000_gen_stvx): Likewise.
26222 (replace_swapped_aligned_store): Likewise.
26223 (rs6000_gen_lvx): Likewise.
26224 (replace_swapped_aligned_load): Likewise.
26225 (replace_swapped_load_constant): Capitalize argument name in
26226 comment describing this function.
26227 (rs6000_analyze_swaps): Add a third pass to search for vector loads
26228 and stores that access quad-word aligned addresses and replace
26229 with stvx or lvx instructions when appropriate.
26230 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
26231 New function prototype.
26232 (rs6000_quadword_masked_address_p): Likewise.
26233 (rs6000_gen_lvx): Likewise.
26234 (rs6000_gen_stvx): Likewise.
26235 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
26236 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
26237 when memory address is aligned.
26238 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
26239 this split to select lvx instruction when memory address is aligned.
26240 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
26241 instruction when memory address is aligned.
26242 (*vsx_le_perm_load_v16qi): Likewise.
26243 (four unnamed splitters): Modify to select the stvx instruction
26244 when memory is aligned.
26246 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
26248 * predict.c (determine_unlikely_bbs): Handle correctly BBs
26249 which appears in the queue multiple times.
26251 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26252 Alan Hayward <alan.hayward@arm.com>
26253 David Sherwood <david.sherwood@arm.com>
26255 * tree-vectorizer.h (vec_lower_bound): New structure.
26256 (_loop_vec_info): Add check_nonzero and lower_bounds.
26257 (LOOP_VINFO_CHECK_NONZERO): New macro.
26258 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
26259 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
26260 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
26261 fields. Make seg_len the distance travelled, not including the
26263 (dr_direction_indicator): Declare.
26264 (dr_zero_step_indicator): Likewise.
26265 (dr_known_forward_stride_p): Likewise.
26266 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
26268 (runtime_alias_check_p): Allow runtime alias checks with
26270 (operator ==): Compare access_size and align.
26271 (prune_runtime_alias_test_list): Rework for new distinction between
26272 the access_size and seg_len.
26273 (create_intersect_range_checks_index): Likewise. Cope with polynomial
26275 (get_segment_min_max): New function.
26276 (create_intersect_range_checks): Use it.
26277 (dr_step_indicator): New function.
26278 (dr_direction_indicator): Likewise.
26279 (dr_zero_step_indicator): Likewise.
26280 (dr_known_forward_stride_p): Likewise.
26281 * tree-loop-distribution.c (data_ref_segment_size): Return
26282 DR_STEP * (niters - 1).
26283 (compute_alias_check_pairs): Update call to the dr_with_seg_len
26285 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
26286 (vect_preserves_scalar_order_p): New function, split out from...
26287 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
26288 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
26289 (vect_vfa_access_size): New function.
26290 (vect_vfa_align): Likewise.
26291 (vect_compile_time_alias): Take access_size_a and access_b arguments.
26292 (dump_lower_bound): New function.
26293 (vect_check_lower_bound): Likewise.
26294 (vect_small_gap_p): Likewise.
26295 (vectorizable_with_step_bound_p): Likewise.
26296 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
26297 depencies if the vectorization factor is 1. Convert the checks
26298 for nonzero steps into checks on the bounds of DR_STEP. Try using
26299 a bunds check for variable steps if the minimum required step is
26300 relatively small. Update calls to the dr_with_seg_len
26301 constructor and to vect_compile_time_alias.
26302 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
26304 (vect_loop_versioning): Call it.
26305 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
26307 (vect_estimate_min_profitable_iters): Account for any bounds checks.
26309 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26310 Alan Hayward <alan.hayward@arm.com>
26311 David Sherwood <david.sherwood@arm.com>
26313 * doc/sourcebuild.texi (vect_scatter_store): Document.
26314 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
26316 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
26318 * genopinit.c (main): Add supports_vec_scatter_store and
26319 supports_vec_scatter_store_cached to target_optabs.
26320 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
26321 IFN_MASK_SCATTER_STORE.
26322 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
26324 * internal-fn.h (internal_store_fn_p): Declare.
26325 (internal_fn_stored_value_index): Likewise.
26326 * internal-fn.c (scatter_store_direct): New macro.
26327 (expand_scatter_store_optab_fn): New function.
26328 (direct_scatter_store_optab_supported_p): New macro.
26329 (internal_store_fn_p): New function.
26330 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
26331 IFN_MASK_SCATTER_STORE.
26332 (internal_fn_mask_index): Likewise.
26333 (internal_fn_stored_value_index): New function.
26334 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
26335 for scatter stores.
26336 * optabs-query.h (supports_vec_scatter_store_p): Declare.
26337 * optabs-query.c (supports_vec_scatter_store_p): New function.
26338 * tree-vectorizer.h (vect_get_store_rhs): Declare.
26339 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
26340 true for scatter stores.
26341 (vect_gather_scatter_fn_p): Handle scatter stores too.
26342 (vect_check_gather_scatter): Consider using scatter stores if
26343 supports_vec_scatter_store_p.
26344 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
26345 scatter stores too.
26346 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
26347 internal_fn_stored_value_index.
26348 (check_load_store_masking): Handle scatter stores too.
26349 (vect_get_store_rhs): Make public.
26350 (vectorizable_call): Use internal_store_fn_p.
26351 (vectorizable_store): Handle scatter store internal functions.
26352 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
26353 when deciding whether the end of the group has been reached.
26354 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
26355 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
26356 (mask_scatter_store<mode>): New insns.
26358 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26359 Alan Hayward <alan.hayward@arm.com>
26360 David Sherwood <david.sherwood@arm.com>
26362 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
26363 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
26364 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
26366 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
26367 Use vect_truncate_gather_scatter_offset if we can't treat the
26368 operation as a normal gather load or scatter store.
26369 (get_group_load_store_type): Take the gather_scatter_info
26370 as argument. Try using a gather load or scatter store for
26371 single-element groups.
26372 (get_load_store_type): Update calls to get_group_load_store_type
26373 and vect_use_strided_gather_scatters_p.
26375 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26376 Alan Hayward <alan.hayward@arm.com>
26377 David Sherwood <david.sherwood@arm.com>
26379 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
26380 optional tree argument.
26381 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
26383 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
26384 but continue to use the current value as a fallback.
26385 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
26386 to compare the updates.
26387 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
26388 (get_load_store_type): Use it when handling a strided access.
26389 (vect_get_strided_load_store_ops): New function.
26390 (vect_get_data_ptr_increment): Likewise.
26391 (vectorizable_load): Handle strided gather loads. Always pass
26392 a step to vect_create_data_ref_ptr and bump_vector_ptr.
26394 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26395 Alan Hayward <alan.hayward@arm.com>
26396 David Sherwood <david.sherwood@arm.com>
26398 * doc/md.texi (gather_load@var{m}): Document.
26399 (mask_gather_load@var{m}): Likewise.
26400 * genopinit.c (main): Add supports_vec_gather_load and
26401 supports_vec_gather_load_cached to target_optabs.
26402 * optabs-tree.c (init_tree_optimization_optabs): Use
26403 ggc_cleared_alloc to allocate target_optabs.
26404 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
26405 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
26407 * internal-fn.h (internal_load_fn_p): Declare.
26408 (internal_gather_scatter_fn_p): Likewise.
26409 (internal_fn_mask_index): Likewise.
26410 (internal_gather_scatter_fn_supported_p): Likewise.
26411 * internal-fn.c (gather_load_direct): New macro.
26412 (expand_gather_load_optab_fn): New function.
26413 (direct_gather_load_optab_supported_p): New macro.
26414 (direct_internal_fn_optab): New function.
26415 (internal_load_fn_p): Likewise.
26416 (internal_gather_scatter_fn_p): Likewise.
26417 (internal_fn_mask_index): Likewise.
26418 (internal_gather_scatter_fn_supported_p): Likewise.
26419 * optabs-query.c (supports_at_least_one_mode_p): New function.
26420 (supports_vec_gather_load_p): Likewise.
26421 * optabs-query.h (supports_vec_gather_load_p): Declare.
26422 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
26423 and memory_type field.
26424 (NUM_PATTERNS): Bump to 15.
26425 * tree-vect-data-refs.c: Include internal-fn.h.
26426 (vect_gather_scatter_fn_p): New function.
26427 (vect_describe_gather_scatter_call): Likewise.
26428 (vect_check_gather_scatter): Try using internal functions for
26429 gather loads. Recognize existing calls to a gather load function.
26430 (vect_analyze_data_refs): Consider using gather loads if
26431 supports_vec_gather_load_p.
26432 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
26433 (vect_get_gather_scatter_offset_type): Likewise.
26434 (vect_convert_mask_for_vectype): Likewise.
26435 (vect_add_conversion_to_patterm): Likewise.
26436 (vect_try_gather_scatter_pattern): Likewise.
26437 (vect_recog_gather_scatter_pattern): New pattern recognizer.
26438 (vect_vect_recog_func_ptrs): Add it.
26439 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
26440 internal_fn_mask_index and internal_gather_scatter_fn_p.
26441 (check_load_store_masking): Take the gather_scatter_info as an
26442 argument and handle gather loads.
26443 (vect_get_gather_scatter_ops): New function.
26444 (vectorizable_call): Check internal_load_fn_p.
26445 (vectorizable_load): Likewise. Handle gather load internal
26447 (vectorizable_store): Update call to check_load_store_masking.
26448 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
26449 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
26450 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
26451 (aarch64_gather_scale_operand_d): New predicates.
26452 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
26453 (mask_gather_load<mode>): New insns.
26455 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26456 Alan Hayward <alan.hayward@arm.com>
26457 David Sherwood <david.sherwood@arm.com>
26459 * optabs.def (fold_left_plus_optab): New optab.
26460 * doc/md.texi (fold_left_plus_@var{m}): Document.
26461 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
26462 * internal-fn.c (fold_left_direct): Define.
26463 (expand_fold_left_optab_fn): Likewise.
26464 (direct_fold_left_optab_supported_p): Likewise.
26465 * fold-const-call.c (fold_const_fold_left): New function.
26466 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
26467 * tree-parloops.c (valid_reduction_p): New function.
26468 (gather_scalar_reductions): Use it.
26469 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
26470 (vect_finish_replace_stmt): Declare.
26471 * tree-vect-loop.c (fold_left_reduction_fn): New function.
26472 (needs_fold_left_reduction_p): New function, split out from...
26473 (vect_is_simple_reduction): ...here. Accept reductions that
26474 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
26475 (vect_force_simple_reduction): Also store the reduction type in
26476 the assignment's STMT_VINFO_REDUC_TYPE.
26477 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
26478 (merge_with_identity): New function.
26479 (vect_expand_fold_left): Likewise.
26480 (vectorize_fold_left_reduction): Likewise.
26481 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
26482 scalar phi in place for it. Check for target support and reject
26483 cases that would reassociate the operation. Defer the transform
26484 phase to vectorize_fold_left_reduction.
26485 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
26486 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
26487 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
26489 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26491 * tree-if-conv.c (predicate_mem_writes): Remove redundant
26492 call to ifc_temp_var.
26494 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26495 Alan Hayward <alan.hayward@arm.com>
26496 David Sherwood <david.sherwood@arm.com>
26498 * target.def (legitimize_address_displacement): Take the original
26499 offset as a poly_int.
26500 * targhooks.h (default_legitimize_address_displacement): Update
26502 * targhooks.c (default_legitimize_address_displacement): Likewise.
26503 * doc/tm.texi: Regenerate.
26504 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
26505 as an argument, moving assert of ad->disp == ad->disp_term to...
26506 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
26507 Try calling targetm.legitimize_address_displacement before expanding
26508 the address rather than afterwards, and adjust for the new interface.
26509 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
26510 Match the new hook interface. Handle SVE addresses.
26511 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
26512 new hook interface.
26514 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26516 * Makefile.in (OBJS): Add early-remat.o.
26517 * target.def (select_early_remat_modes): New hook.
26518 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
26519 * doc/tm.texi: Regenerate.
26520 * targhooks.h (default_select_early_remat_modes): Declare.
26521 * targhooks.c (default_select_early_remat_modes): New function.
26522 * timevar.def (TV_EARLY_REMAT): New timevar.
26523 * passes.def (pass_early_remat): New pass.
26524 * tree-pass.h (make_pass_early_remat): Declare.
26525 * early-remat.c: New file.
26526 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
26528 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
26530 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26531 Alan Hayward <alan.hayward@arm.com>
26532 David Sherwood <david.sherwood@arm.com>
26534 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
26535 vfm1 with a bound_epilog parameter.
26536 (vect_do_peeling): Update calls accordingly, and move the prologue
26537 call earlier in the function. Treat the base bound_epilog as 0 for
26538 fully-masked loops and retain vf - 1 for other loops. Add 1 to
26539 this base when peeling for gaps.
26540 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
26541 with fully-masked loops.
26542 (vect_estimate_min_profitable_iters): Handle the single peeled
26543 iteration in that case.
26545 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26546 Alan Hayward <alan.hayward@arm.com>
26547 David Sherwood <david.sherwood@arm.com>
26549 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
26550 single-element interleaving even if the size is not a power of 2.
26551 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
26552 accesses for single-element interleaving if the group size is
26555 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26556 Alan Hayward <alan.hayward@arm.com>
26557 David Sherwood <david.sherwood@arm.com>
26559 * doc/md.texi (fold_extract_last_@var{m}): Document.
26560 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
26561 * optabs.def (fold_extract_last_optab): New optab.
26562 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
26563 * internal-fn.c (fold_extract_direct): New macro.
26564 (expand_fold_extract_optab_fn): Likewise.
26565 (direct_fold_extract_optab_supported_p): Likewise.
26566 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
26567 * tree-vect-loop.c (vect_model_reduction_cost): Handle
26568 EXTRACT_LAST_REDUCTION.
26569 (get_initial_def_for_reduction): Do not create an initial vector
26570 for EXTRACT_LAST_REDUCTION reductions.
26571 (vectorizable_reduction): Leave the scalar phi in place for
26572 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
26573 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
26574 epilogue code for EXTRACT_LAST_REDUCTION and defer the
26575 transform phase to vectorizable_condition.
26576 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
26578 (vect_finish_stmt_generation): ...here.
26579 (vect_finish_replace_stmt): New function.
26580 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
26581 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
26583 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
26585 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26586 Alan Hayward <alan.hayward@arm.com>
26587 David Sherwood <david.sherwood@arm.com>
26589 * doc/md.texi (extract_last_@var{m}): Document.
26590 * optabs.def (extract_last_optab): New optab.
26591 * internal-fn.def (EXTRACT_LAST): New internal function.
26592 * internal-fn.c (cond_unary_direct): New macro.
26593 (expand_cond_unary_optab_fn): Likewise.
26594 (direct_cond_unary_optab_supported_p): Likewise.
26595 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
26596 loops using EXTRACT_LAST.
26597 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
26598 (extract_last_<mode>): ...this optab.
26599 (vec_extract<mode><Vel>): Update accordingly.
26601 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26602 Alan Hayward <alan.hayward@arm.com>
26603 David Sherwood <david.sherwood@arm.com>
26605 * target.def (empty_mask_is_expensive): New hook.
26606 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
26607 * doc/tm.texi: Regenerate.
26608 * targhooks.h (default_empty_mask_is_expensive): Declare.
26609 * targhooks.c (default_empty_mask_is_expensive): New function.
26610 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
26611 if the target says that empty masks are expensive.
26612 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
26614 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
26616 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26617 Alan Hayward <alan.hayward@arm.com>
26618 David Sherwood <david.sherwood@arm.com>
26620 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
26621 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
26622 (vect_use_loop_mask_for_alignment_p): New function.
26623 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
26624 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
26625 niters_skip argument. Make sure that the first niters_skip elements
26626 of the first iteration are inactive.
26627 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
26628 Update call to vect_set_loop_masks_directly.
26629 (get_misalign_in_elems): New function, split out from...
26630 (vect_gen_prolog_loop_niters): ...here.
26631 (vect_update_init_of_dr): Take a code argument that specifies whether
26632 the adjustment should be added or subtracted.
26633 (vect_update_init_of_drs): Likewise.
26634 (vect_prepare_for_masked_peels): New function.
26635 (vect_do_peeling): Skip prologue peeling if we're using a mask
26636 instead. Update call to vect_update_inits_of_drs.
26637 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
26639 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
26640 alignment. Do not include the number of peeled iterations in
26641 the minimum threshold in that case.
26642 (vectorizable_induction): Adjust the start value down by
26643 LOOP_VINFO_MASK_SKIP_NITERS iterations.
26644 (vect_transform_loop): Call vect_prepare_for_masked_peels.
26645 Take the number of skipped iterations into account when calculating
26647 * tree-vect-stmts.c (vect_gen_while_not): New function.
26649 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26650 Alan Hayward <alan.hayward@arm.com>
26651 David Sherwood <david.sherwood@arm.com>
26653 * doc/sourcebuild.texi (vect_fully_masked): Document.
26654 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
26655 default value to 0.
26656 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
26658 (vect_analyze_loop_2): ...here. Don't check the vectorization
26659 factor against the number of loop iterations if the loop is
26662 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26663 Alan Hayward <alan.hayward@arm.com>
26664 David Sherwood <david.sherwood@arm.com>
26666 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
26667 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
26668 (dump_groups): Update accordingly.
26669 (iv_use::mem_type): New member variable.
26670 (address_p): New function.
26671 (record_use): Add a mem_type argument and initialize the new
26673 (record_group_use): Add a mem_type argument. Use address_p.
26674 Remove obsolete null checks of base_object. Update call to record_use.
26675 (find_interesting_uses_op): Update call to record_group_use.
26676 (find_interesting_uses_cond): Likewise.
26677 (find_interesting_uses_address): Likewise.
26678 (get_mem_type_for_internal_fn): New function.
26679 (find_address_like_use): Likewise.
26680 (find_interesting_uses_stmt): Try find_address_like_use before
26681 calling find_interesting_uses_op.
26682 (addr_offset_valid_p): Use the iv mem_type field as the type
26683 of the addressed memory.
26684 (add_autoinc_candidates): Likewise.
26685 (get_address_cost): Likewise.
26686 (split_small_address_groups_p): Use address_p.
26687 (split_address_groups): Likewise.
26688 (add_iv_candidate_for_use): Likewise.
26689 (autoinc_possible_for_pair): Likewise.
26690 (rewrite_groups): Likewise.
26691 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
26692 (determine_group_iv_cost): Update after split of USE_ADDRESS.
26693 (get_alias_ptr_type_for_ptr_address): New function.
26694 (rewrite_use_address): Rewrite address uses in calls that were
26695 identified by find_address_like_use.
26697 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26698 Alan Hayward <alan.hayward@arm.com>
26699 David Sherwood <david.sherwood@arm.com>
26701 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
26703 * gimple-expr.h (is_gimple_addressable: Likewise.
26704 * gimple-expr.c (is_gimple_address): Likewise.
26705 * internal-fn.c (expand_call_mem_ref): New function.
26706 (expand_mask_load_optab_fn): Use it.
26707 (expand_mask_store_optab_fn): Likewise.
26709 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26710 Alan Hayward <alan.hayward@arm.com>
26711 David Sherwood <david.sherwood@arm.com>
26713 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
26714 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
26715 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
26716 (cond_umax@var{mode}): Document.
26717 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
26718 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
26719 (cond_umin_optab, cond_umax_optab): New optabs.
26720 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
26721 (COND_IOR, COND_XOR): New internal functions.
26722 * internal-fn.h (get_conditional_internal_fn): Declare.
26723 * internal-fn.c (cond_binary_direct): New macro.
26724 (expand_cond_binary_optab_fn): Likewise.
26725 (direct_cond_binary_optab_supported_p): Likewise.
26726 (get_conditional_internal_fn): New function.
26727 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
26728 Cope with reduction statements that are vectorized as calls rather
26730 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
26731 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
26732 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
26733 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
26734 (UNSPEC_COND_EOR): New unspecs.
26735 (optab): Add mappings for them.
26736 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
26737 (sve_int_op, sve_fp_op): New int attributes.
26739 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26740 Alan Hayward <alan.hayward@arm.com>
26741 David Sherwood <david.sherwood@arm.com>
26743 * optabs.def (while_ult_optab): New optab.
26744 * doc/md.texi (while_ult@var{m}@var{n}): Document.
26745 * internal-fn.def (WHILE_ULT): New internal function.
26746 * internal-fn.h (direct_internal_fn_supported_p): New override
26747 that takes two types as argument.
26748 * internal-fn.c (while_direct): New macro.
26749 (expand_while_optab_fn): New function.
26750 (convert_optab_supported_p): Likewise.
26751 (direct_while_optab_supported_p): New macro.
26752 * wide-int.h (wi::udiv_ceil): New function.
26753 * tree-vectorizer.h (rgroup_masks): New structure.
26754 (vec_loop_masks): New typedef.
26755 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
26756 and fully_masked_p.
26757 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
26758 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
26759 (vect_max_vf): New function.
26760 (slpeel_make_loop_iterate_ntimes): Delete.
26761 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
26762 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
26763 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
26764 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
26765 internal-fn.h, stor-layout.h and optabs-query.h.
26766 (vect_set_loop_mask): New function.
26767 (add_preheader_seq): Likewise.
26768 (add_header_seq): Likewise.
26769 (interleave_supported_p): Likewise.
26770 (vect_maybe_permute_loop_masks): Likewise.
26771 (vect_set_loop_masks_directly): Likewise.
26772 (vect_set_loop_condition_masked): Likewise.
26773 (vect_set_loop_condition_unmasked): New function, split out from
26774 slpeel_make_loop_iterate_ntimes.
26775 (slpeel_make_loop_iterate_ntimes): Rename to..
26776 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
26777 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
26778 (vect_do_peeling): Update call accordingly.
26779 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
26781 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
26782 mask_compare_type, can_fully_mask_p and fully_masked_p.
26783 (release_vec_loop_masks): New function.
26784 (_loop_vec_info): Use it to free the loop masks.
26785 (can_produce_all_loop_masks_p): New function.
26786 (vect_get_max_nscalars_per_iter): Likewise.
26787 (vect_verify_full_masking): Likewise.
26788 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
26789 retries, and free the mask rgroups before retrying. Check loop-wide
26790 reasons for disallowing fully-masked loops. Make the final decision
26791 about whether use a fully-masked loop or not.
26792 (vect_estimate_min_profitable_iters): Do not assume that peeling
26793 for the number of iterations will be needed for fully-masked loops.
26794 (vectorizable_reduction): Disable fully-masked loops.
26795 (vectorizable_live_operation): Likewise.
26796 (vect_halve_mask_nunits): New function.
26797 (vect_double_mask_nunits): Likewise.
26798 (vect_record_loop_mask): Likewise.
26799 (vect_get_loop_mask): Likewise.
26800 (vect_transform_loop): Handle the case in which the final loop
26801 iteration might handle a partial vector. Call vect_set_loop_condition
26802 instead of slpeel_make_loop_iterate_ntimes.
26803 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
26804 (check_load_store_masking): New function.
26805 (prepare_load_store_mask): Likewise.
26806 (vectorizable_store): Handle fully-masked loops.
26807 (vectorizable_load): Likewise.
26808 (supportable_widening_operation): Use vect_halve_mask_nunits for
26810 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
26811 (vect_gen_while): New function.
26812 * config/aarch64/aarch64.md (umax<mode>3): New expander.
26813 (aarch64_uqdec<mode>): New insn.
26815 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26816 Alan Hayward <alan.hayward@arm.com>
26817 David Sherwood <david.sherwood@arm.com>
26819 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
26820 (reduc_xor_scal_optab): New optabs.
26821 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
26822 (reduc_xor_scal_@var{m}): Document.
26823 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
26824 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
26825 internal functions.
26826 * fold-const-call.c (fold_const_call): Handle them.
26827 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
26828 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
26829 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
26830 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
26831 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
26832 (UNSPEC_XORV): New unspecs.
26833 (optab): Add entries for them.
26834 (BITWISEV): New int iterator.
26835 (bit_reduc_op): New int attributes.
26837 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26838 Alan Hayward <alan.hayward@arm.com>
26839 David Sherwood <david.sherwood@arm.com>
26841 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
26842 * internal-fn.def (VEC_SHL_INSERT): New internal function.
26843 * optabs.def (vec_shl_insert_optab): New optab.
26844 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
26845 (duplicate_and_interleave): Likewise.
26846 * tree-vect-loop.c: Include internal-fn.h.
26847 (neutral_op_for_slp_reduction): New function, split out from
26848 get_initial_defs_for_reduction.
26849 (get_initial_def_for_reduction): Handle option 2 for variable-length
26850 vectors by loading the neutral value into a vector and then shifting
26851 the initial value into element 0.
26852 (get_initial_defs_for_reduction): Replace the code argument with
26853 the neutral value calculated by neutral_op_for_slp_reduction.
26854 Use gimple_build_vector for constant-length vectors.
26855 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
26856 but the first group_size elements have a neutral value.
26857 Use duplicate_and_interleave otherwise.
26858 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
26859 Update call to get_initial_defs_for_reduction. Handle SLP
26860 reductions for variable-length vectors by creating one vector
26861 result for each scalar result, with the elements associated
26862 with other scalar results stubbed out with the neutral value.
26863 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
26864 Require IFN_VEC_SHL_INSERT for double reductions on
26865 variable-length vectors, or SLP reductions that have
26866 a neutral value. Require can_duplicate_and_interleave_p
26867 support for variable-length unchained SLP reductions if there
26868 is no neutral value, such as for MIN/MAX reductions. Also require
26869 the number of vector elements to be a multiple of the number of
26870 SLP statements when doing variable-length unchained SLP reductions.
26871 Update call to vect_create_epilog_for_reduction.
26872 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
26873 and remove initial values.
26874 (duplicate_and_interleave): Make public.
26875 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
26876 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
26878 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26879 Alan Hayward <alan.hayward@arm.com>
26880 David Sherwood <david.sherwood@arm.com>
26882 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
26883 (can_duplicate_and_interleave_p): New function.
26884 (vect_get_and_check_slp_defs): Take the vector of statements
26885 rather than just the current one. Remove excess parentheses.
26886 Restriction rejectinon of vect_constant_def and vect_external_def
26887 for variable-length vectors to boolean types, or types for which
26888 can_duplicate_and_interleave_p is false.
26889 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
26890 (duplicate_and_interleave): New function.
26891 (vect_get_constant_vectors): Use gimple_build_vector for
26892 constant-length vectors and suitable variable-length constant
26893 vectors. Use duplicate_and_interleave for other variable-length
26894 vectors. Don't defer the update when inserting new statements.
26896 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26897 Alan Hayward <alan.hayward@arm.com>
26898 David Sherwood <david.sherwood@arm.com>
26900 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
26901 min_profitable_iters doesn't go negative.
26903 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26904 Alan Hayward <alan.hayward@arm.com>
26905 David Sherwood <david.sherwood@arm.com>
26907 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
26908 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
26909 * optabs.def (vec_mask_load_lanes_optab): New optab.
26910 (vec_mask_store_lanes_optab): Likewise.
26911 * internal-fn.def (MASK_LOAD_LANES): New internal function.
26912 (MASK_STORE_LANES): Likewise.
26913 * internal-fn.c (mask_load_lanes_direct): New macro.
26914 (mask_store_lanes_direct): Likewise.
26915 (expand_mask_load_optab_fn): Handle masked operations.
26916 (expand_mask_load_lanes_optab_fn): New macro.
26917 (expand_mask_store_optab_fn): Handle masked operations.
26918 (expand_mask_store_lanes_optab_fn): New macro.
26919 (direct_mask_load_lanes_optab_supported_p): Likewise.
26920 (direct_mask_store_lanes_optab_supported_p): Likewise.
26921 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
26923 (vect_load_lanes_supported): Likewise.
26924 * tree-vect-data-refs.c (strip_conversion): New function.
26925 (can_group_stmts_p): Likewise.
26926 (vect_analyze_data_ref_accesses): Use it instead of checking
26927 for a pair of assignments.
26928 (vect_store_lanes_supported): Take a masked_p parameter.
26929 (vect_load_lanes_supported): Likewise.
26930 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
26931 vect_store_lanes_supported and vect_load_lanes_supported.
26932 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
26933 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
26934 parameter. Don't allow gaps for masked accesses.
26935 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
26936 and vect_load_lanes_supported.
26937 (get_load_store_type): Take a masked_p parameter and update
26938 call to get_group_load_store_type.
26939 (vectorizable_store): Update call to get_load_store_type.
26940 Handle IFN_MASK_STORE_LANES.
26941 (vectorizable_load): Update call to get_load_store_type.
26942 Handle IFN_MASK_LOAD_LANES.
26944 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26945 Alan Hayward <alan.hayward@arm.com>
26946 David Sherwood <david.sherwood@arm.com>
26948 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
26950 * config/aarch64/aarch64-protos.h
26951 (aarch64_sve_struct_memory_operand_p): Declare.
26952 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
26953 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
26954 (VPRED, vpred): Handle SVE structure modes.
26955 * config/aarch64/constraints.md (Utx): New constraint.
26956 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
26957 (aarch64_sve_struct_nonimmediate_operand): New predicates.
26958 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
26959 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
26960 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
26961 structure modes. Split into pieces after RA.
26962 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
26963 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
26965 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
26966 SVE structure modes.
26967 (aarch64_classify_address): Likewise.
26968 (sizetochar): Move earlier in file.
26969 (aarch64_print_operand): Handle SVE register lists.
26970 (aarch64_array_mode): New function.
26971 (aarch64_sve_struct_memory_operand_p): Likewise.
26972 (TARGET_ARRAY_MODE): Redefine.
26974 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26975 Alan Hayward <alan.hayward@arm.com>
26976 David Sherwood <david.sherwood@arm.com>
26978 * target.def (array_mode): New target hook.
26979 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
26980 * doc/tm.texi: Regenerate.
26981 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
26982 * hooks.c (hook_optmode_mode_uhwi_none): New function.
26983 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
26984 targetm.array_mode.
26985 * stor-layout.c (mode_for_array): Likewise. Support polynomial
26988 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26989 Alan Hayward <alan.hayward@arm.com>
26990 David Sherwood <david.sherwood@arm.com>
26992 * fold-const.c (fold_binary_loc): Check the argument types
26993 rather than the result type when testing for a vector operation.
26995 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26997 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
26998 * doc/tm.texi: Regenerate.
27000 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27001 Alan Hayward <alan.hayward@arm.com>
27002 David Sherwood <david.sherwood@arm.com>
27004 * doc/invoke.texi (-msve-vector-bits=): Document new option.
27005 (sve): Document new AArch64 extension.
27006 * doc/md.texi (w): Extend the description of the AArch64
27007 constraint to include SVE vectors.
27008 (Upl, Upa): Document new AArch64 predicate constraints.
27009 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
27011 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
27012 (msve-vector-bits=): New option.
27013 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
27014 SVE when these are disabled.
27015 (sve): New extension.
27016 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
27017 modes. Adjust their number of units based on aarch64_sve_vg.
27018 (MAX_BITSIZE_MODE_ANY_MODE): Define.
27019 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
27020 aarch64_addr_query_type.
27021 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
27022 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
27023 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
27024 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
27025 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
27026 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
27027 (aarch64_simd_imm_zero_p): Delete.
27028 (aarch64_check_zero_based_sve_index_immediate): Declare.
27029 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
27030 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
27031 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
27032 (aarch64_sve_float_mul_immediate_p): Likewise.
27033 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
27034 rather than an rtx.
27035 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
27036 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
27037 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
27038 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
27039 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
27040 (aarch64_regmode_natural_size): Likewise.
27041 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
27042 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
27044 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
27045 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
27046 for VG and the SVE predicate registers.
27047 (V_ALIASES): Add a "z"-prefixed alias.
27048 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
27049 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
27050 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
27051 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
27052 (REG_CLASS_NAMES): Add entries for them.
27053 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
27054 and the predicate registers.
27055 (aarch64_sve_vg): Declare.
27056 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
27057 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
27058 (REGMODE_NATURAL_SIZE): Define.
27059 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
27061 * config/aarch64/aarch64.c: Include cfgrtl.h.
27062 (simd_immediate_info): Add a constructor for series vectors,
27063 and an associated step field.
27064 (aarch64_sve_vg): New variable.
27065 (aarch64_dbx_register_number): Handle VG and the predicate registers.
27066 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
27067 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
27068 (VEC_ANY_DATA, VEC_STRUCT): New constants.
27069 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
27070 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
27071 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
27072 (aarch64_get_mask_mode): New functions.
27073 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
27074 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
27075 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
27076 predicate modes and predicate registers. Explicitly restrict
27077 GPRs to modes of 16 bytes or smaller. Only allow FP registers
27078 to store a vector mode if it is recognized by
27079 aarch64_classify_vector_mode.
27080 (aarch64_regmode_natural_size): New function.
27081 (aarch64_hard_regno_caller_save_mode): Return the original mode
27083 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
27084 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
27085 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
27086 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
27088 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
27089 does not overlap dest if the function is frame-related. Handle
27091 (aarch64_split_add_offset): New function.
27092 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
27093 them aarch64_add_offset.
27094 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
27095 and update call to aarch64_sub_sp.
27096 (aarch64_add_cfa_expression): New function.
27097 (aarch64_expand_prologue): Pass extra temporary registers to the
27098 functions above. Handle the case in which we need to emit new
27099 DW_CFA_expressions for registers that were originally saved
27100 relative to the stack pointer, but now have to be expressed
27101 relative to the frame pointer.
27102 (aarch64_output_mi_thunk): Pass extra temporary registers to the
27104 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
27105 IP0 and IP1 values for SVE frames.
27106 (aarch64_expand_vec_series): New function.
27107 (aarch64_expand_sve_widened_duplicate): Likewise.
27108 (aarch64_expand_sve_const_vector): Likewise.
27109 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
27110 Handle SVE constants. Use emit_move_insn to move a force_const_mem
27111 into the register, rather than emitting a SET directly.
27112 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
27113 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
27114 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
27115 (offset_9bit_signed_scaled_p): New functions.
27116 (aarch64_replicate_bitmask_imm): New function.
27117 (aarch64_bitmask_imm): Use it.
27118 (aarch64_cannot_force_const_mem): Reject expressions involving
27119 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
27120 (aarch64_classify_index): Handle SVE indices, by requiring
27121 a plain register index with a scale that matches the element size.
27122 (aarch64_classify_address): Handle SVE addresses. Assert that
27123 the mode of the address is VOIDmode or an integer mode.
27124 Update call to aarch64_classify_symbol.
27125 (aarch64_classify_symbolic_expression): Update call to
27126 aarch64_classify_symbol.
27127 (aarch64_const_vec_all_in_range_p): New function.
27128 (aarch64_print_vector_float_operand): Likewise.
27129 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
27130 "vN" for FP registers with SVE modes. Handle (const ...) vectors
27131 and the FP immediates 1.0 and 0.5.
27132 (aarch64_print_address_internal): Handle SVE addresses.
27133 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
27134 (aarch64_regno_regclass): Handle predicate registers.
27135 (aarch64_secondary_reload): Handle big-endian reloads of SVE
27137 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
27138 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
27139 (aarch64_convert_sve_vector_bits): New function.
27140 (aarch64_override_options): Use it to handle -msve-vector-bits=.
27141 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
27142 rather than an rtx.
27143 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
27144 Handle SVE vector and predicate modes. Accept VL-based constants
27145 that need only one temporary register, and VL offsets that require
27146 no temporary registers.
27147 (aarch64_conditional_register_usage): Mark the predicate registers
27148 as fixed if SVE isn't available.
27149 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
27150 Return true for SVE vector and predicate modes.
27151 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
27152 rather than an unsigned int. Handle SVE modes.
27153 (aarch64_preferred_simd_mode): Update call accordingly. Handle
27155 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
27157 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
27158 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
27159 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
27160 (aarch64_sve_float_mul_immediate_p): New functions.
27161 (aarch64_sve_valid_immediate): New function.
27162 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
27163 Explicitly reject structure modes. Check for INDEX constants.
27164 Handle PTRUE and PFALSE constants.
27165 (aarch64_check_zero_based_sve_index_immediate): New function.
27166 (aarch64_simd_imm_zero_p): Delete.
27167 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
27168 vector modes. Accept constants in the range of CNT[BHWD].
27169 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
27170 ask for an Advanced SIMD mode.
27171 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
27172 (aarch64_simd_vector_alignment): Handle SVE predicates.
27173 (aarch64_vectorize_preferred_vector_alignment): New function.
27174 (aarch64_simd_vector_alignment_reachable): Use it instead of
27176 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
27177 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
27179 (MAX_VECT_LEN): Delete.
27180 (expand_vec_perm_d): Add a vec_flags field.
27181 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
27182 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
27183 (aarch64_evpc_ext): Don't apply a big-endian lane correction
27185 (aarch64_evpc_rev): Rename to...
27186 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
27187 (aarch64_evpc_rev_global): New function.
27188 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
27189 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
27191 (aarch64_evpc_sve_tbl): New function.
27192 (aarch64_expand_vec_perm_const_1): Update after rename of
27193 aarch64_evpc_rev. Handle SVE permutes too, trying
27194 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
27195 than aarch64_evpc_tbl.
27196 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
27197 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
27198 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
27199 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
27200 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
27201 (aarch64_expand_sve_vcond): New functions.
27202 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
27203 of aarch64_vector_mode_p.
27204 (aarch64_dwarf_poly_indeterminate_value): New function.
27205 (aarch64_compute_pressure_classes): Likewise.
27206 (aarch64_can_change_mode_class): Likewise.
27207 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
27208 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
27209 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
27210 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
27211 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
27212 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
27213 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
27214 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
27216 (Dn, Dl, Dr): Accept const as well as const_vector.
27217 (Dz): Likewise. Compare against CONST0_RTX.
27218 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
27219 of "vector" where appropriate.
27220 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
27221 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
27222 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
27223 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
27224 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
27225 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
27226 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
27227 (v_int_equiv): Extend to SVE modes.
27228 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
27230 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
27231 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
27232 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
27233 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
27234 (SVE_COND_FP_CMP): New int iterators.
27235 (perm_hilo): Handle the new unpack unspecs.
27236 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
27238 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
27239 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
27240 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
27241 (aarch64_equality_operator, aarch64_constant_vector_operand)
27242 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
27243 (aarch64_sve_nonimmediate_operand): Likewise.
27244 (aarch64_sve_general_operand): Likewise.
27245 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
27246 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
27247 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
27248 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
27249 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
27250 (aarch64_sve_float_arith_immediate): Likewise.
27251 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
27252 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
27253 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
27254 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
27255 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
27256 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
27257 (aarch64_sve_float_arith_operand): Likewise.
27258 (aarch64_sve_float_arith_with_sub_operand): Likewise.
27259 (aarch64_sve_float_mul_operand): Likewise.
27260 (aarch64_sve_vec_perm_operand): Likewise.
27261 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
27262 (aarch64_mov_operand): Accept const_poly_int and const_vector.
27263 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
27264 as well as const_vector.
27265 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
27266 in file. Use CONST0_RTX and CONSTM1_RTX.
27267 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
27268 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
27269 Use aarch64_simd_imm_zero.
27270 * config/aarch64/aarch64-sve.md: New file.
27271 * config/aarch64/aarch64.md: Include it.
27272 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
27273 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
27274 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
27275 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
27276 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
27277 (sve): New attribute.
27278 (enabled): Disable instructions with the sve attribute unless
27280 (movqi, movhi): Pass CONST_POLY_INT operaneds through
27281 aarch64_expand_mov_immediate.
27282 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
27283 CNT[BHSD] immediates.
27284 (movti): Split CONST_POLY_INT moves into two halves.
27285 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
27286 Split additions that need a temporary here if the destination
27287 is the stack pointer.
27288 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
27289 (*add<mode>3_poly_1): New instruction.
27290 (set_clobber_cc): New expander.
27292 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27294 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
27295 parameter and use it instead of GET_MODE_SIZE (innermode). Use
27296 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
27297 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
27298 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
27299 Change innermode from fixed_mode_size to machine_mode.
27300 (simplify_subreg): Update call accordingly. Handle a constant-sized
27301 subreg of a variable-length CONST_VECTOR.
27303 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27304 Alan Hayward <alan.hayward@arm.com>
27305 David Sherwood <david.sherwood@arm.com>
27307 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
27308 (add_offset_to_base): New function, split out from...
27309 (create_mem_ref): ...here. When handling a scale other than 1,
27310 check first whether the address is valid without the offset.
27311 Add it into the base if so, leaving the index and scale as-is.
27313 2018-01-12 Jakub Jelinek <jakub@redhat.com>
27316 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
27317 fold_for_warn before checking if arg2 is INTEGER_CST.
27319 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
27321 * config/rs6000/predicates.md (load_multiple_operation): Delete.
27322 (store_multiple_operation): Delete.
27323 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
27324 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
27325 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
27326 guarded by TARGET_STRING.
27327 (rs6000_output_load_multiple): Delete.
27328 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27329 OPTION_MASK_STRING / TARGET_STRING handling.
27330 (print_operand) <'N', 'O'>: Add comment that these are unused now.
27331 (const rs6000_opt_masks) <"string">: Change mask to 0.
27332 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
27333 (MASK_STRING): Delete.
27334 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
27336 (load_multiple): Delete.
27343 (store_multiple): Delete.
27350 (movmemsi_8reg): Delete.
27351 (corresponding unnamed define_insn): Delete.
27352 (movmemsi_6reg): Delete.
27353 (corresponding unnamed define_insn): Delete.
27354 (movmemsi_4reg): Delete.
27355 (corresponding unnamed define_insn): Delete.
27356 (movmemsi_2reg): Delete.
27357 (corresponding unnamed define_insn): Delete.
27358 (movmemsi_1reg): Delete.
27359 (corresponding unnamed define_insn): Delete.
27360 * config/rs6000/rs6000.opt (mno-string): New.
27361 (mstring): Replace by deprecation warning stub.
27362 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
27364 2018-01-12 Jakub Jelinek <jakub@redhat.com>
27366 * regrename.c (regrename_do_replace): If replacing the same
27367 reg multiple times, try to reuse last created gen_raw_REG.
27370 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
27371 main to workaround a bug in GDB.
27373 2018-01-12 Tom de Vries <tom@codesourcery.com>
27376 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
27378 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
27380 PR rtl-optimization/80481
27381 * ira-color.c (get_cap_member): New function.
27382 (allocnos_conflict_by_live_ranges_p): Use it.
27383 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
27384 (setup_slot_coalesced_allocno_live_ranges): Ditto.
27386 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
27389 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
27390 (*saddl_se_1): Ditto.
27391 (*ssubsi_1): Ditto.
27392 (*ssubl_se_1): Ditto.
27394 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
27396 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
27397 rather than wi::to_widest for DR_INITs.
27398 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
27399 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
27400 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
27402 (vect_analyze_group_access_1): Note that here.
27404 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
27406 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
27407 polynomial type sizes.
27409 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
27411 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
27412 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
27413 (gimple_add_tmp_var): Likewise.
27415 2018-01-12 Martin Liska <mliska@suse.cz>
27417 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
27418 (gimple_alloc_sizes): Likewise.
27419 (dump_gimple_statistics): Use PRIu64 in printf format.
27420 * gimple.h: Change uint64_t to int.
27422 2018-01-12 Martin Liska <mliska@suse.cz>
27424 * tree-core.h: Use uint64_t instead of int.
27425 * tree.c (tree_node_counts): Likewise.
27426 (tree_node_sizes): Likewise.
27427 (dump_tree_statistics): Use PRIu64 in printf format.
27429 2018-01-12 Martin Liska <mliska@suse.cz>
27431 * Makefile.in: As qsort_chk is implemented in vec.c, add
27432 vec.o to linkage of gencfn-macros.
27433 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
27434 passing the info to record_node_allocation_statistics.
27435 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
27437 * ggc-common.c (struct ggc_usage): Add operator== and use
27438 it in operator< and compare function.
27439 * mem-stats.h (struct mem_usage): Likewise.
27440 * vec.c (struct vec_usage): Remove operator< and compare
27441 function. Can be simply inherited.
27443 2018-01-12 Martin Jambor <mjambor@suse.cz>
27446 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
27447 * tree-ssa-math-opts.c: Include domwalk.h.
27448 (convert_mult_to_fma_1): New function.
27449 (fma_transformation_info): New type.
27450 (fma_deferring_state): Likewise.
27451 (cancel_fma_deferring): New function.
27452 (result_of_phi): Likewise.
27453 (last_fma_candidate_feeds_initial_phi): Likewise.
27454 (convert_mult_to_fma): Added deferring logic, split actual
27455 transformation to convert_mult_to_fma_1.
27456 (math_opts_dom_walker): New type.
27457 (math_opts_dom_walker::after_dom_children): New method, body moved
27458 here from pass_optimize_widening_mul::execute, added deferring logic
27460 (pass_optimize_widening_mul::execute): Moved most of code to
27461 math_opts_dom_walker::after_dom_children.
27462 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
27463 * config/i386/i386.c (ix86_option_override_internal): Added
27464 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
27466 2018-01-12 Richard Biener <rguenther@suse.de>
27469 * dwarf2out.c (gen_variable_die): Do not reset old_die for
27470 inline instance vars.
27472 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
27475 * config/rx/rx.c (rx_is_restricted_memory_address):
27476 Handle SUBREG case.
27478 2018-01-12 Richard Biener <rguenther@suse.de>
27480 PR tree-optimization/80846
27481 * target.def (split_reduction): New target hook.
27482 * targhooks.c (default_split_reduction): New function.
27483 * targhooks.h (default_split_reduction): Declare.
27484 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
27485 target requests first reduce vectors by combining low and high
27487 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
27488 (get_vectype_for_scalar_type_and_size): Export.
27489 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
27490 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
27491 * doc/tm.texi: Regenerate.
27492 * config/i386/i386.c (ix86_split_reduction): Implement
27493 TARGET_VECTORIZE_SPLIT_REDUCTION.
27495 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
27498 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
27499 in PIC mode except for TARGET_VXWORKS_RTP.
27500 * config/sparc/sparc.c: Include cfgrtl.h.
27501 (TARGET_INIT_PIC_REG): Define.
27502 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
27503 (sparc_pic_register_p): New predicate.
27504 (sparc_legitimate_address_p): Use it.
27505 (sparc_legitimize_pic_address): Likewise.
27506 (sparc_delegitimize_address): Likewise.
27507 (sparc_mode_dependent_address_p): Likewise.
27508 (gen_load_pcrel_sym): Remove 4th parameter.
27509 (load_got_register): Adjust call to above. Remove obsolete stuff.
27510 (sparc_expand_prologue): Do not call load_got_register here.
27511 (sparc_flat_expand_prologue): Likewise.
27512 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
27513 (sparc_use_pseudo_pic_reg): New function.
27514 (sparc_init_pic_reg): Likewise.
27515 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
27516 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
27518 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
27520 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27521 Add item for branch_cost.
27523 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
27525 PR rtl-optimization/83565
27526 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
27527 not extend the result to a larger mode for rotate operations.
27528 (num_sign_bit_copies1): Likewise.
27530 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27533 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
27535 Use values-Xc.o for -pedantic.
27536 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
27538 2018-01-12 Martin Liska <mliska@suse.cz>
27541 * ipa-devirt.c (final_warning_record::grow_type_warnings):
27543 (possible_polymorphic_call_targets): Use it.
27544 (ipa_devirt): Likewise.
27546 2018-01-12 Martin Liska <mliska@suse.cz>
27548 * profile-count.h (enum profile_quality): Use 0 as invalid
27549 enum value of profile_quality.
27551 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
27553 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
27554 -mext-string options.
27556 2018-01-12 Richard Biener <rguenther@suse.de>
27558 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
27559 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
27560 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
27562 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
27564 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
27566 * configure.ac (--with-long-double-format): Add support for the
27567 configuration option to change the default long double format on
27569 * config.gcc (powerpc*-linux*-*): Likewise.
27570 * configure: Regenerate.
27571 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
27572 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
27573 used without modification.
27575 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27577 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
27578 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
27579 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
27580 MISC_BUILTIN_SPEC_BARRIER.
27581 (rs6000_init_builtins): Likewise.
27582 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
27584 (speculation_barrier): New define_insn.
27585 * doc/extend.texi: Document __builtin_speculation_barrier.
27587 2018-01-11 Jakub Jelinek <jakub@redhat.com>
27590 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
27591 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
27592 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
27594 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
27595 integral modes instead of "ss" and "sd".
27596 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
27597 vectors with 32-bit and 64-bit elements.
27598 (vecdupssescalarmodesuffix): New mode attribute.
27599 (vec_dup<mode>): Use it.
27601 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
27604 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
27605 frame if argument is passed on stack.
27607 2018-01-11 Jakub Jelinek <jakub@redhat.com>
27610 * ree.c (combine_reaching_defs): Optimize also
27611 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
27612 reg2=any_extend(exp); reg1=reg2;, formatting fix.
27614 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
27616 PR middle-end/83189
27617 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
27619 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
27621 PR middle-end/83718
27622 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
27623 after they are computed.
27625 2018-01-11 Bin Cheng <bin.cheng@arm.com>
27627 PR tree-optimization/83695
27628 * gimple-loop-linterchange.cc
27629 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
27630 reset cached scev information after interchange.
27631 (pass_linterchange::execute): Remove call to scev_reset_htab.
27633 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27635 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
27636 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
27637 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
27638 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
27639 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
27640 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
27641 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
27642 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
27643 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
27644 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
27645 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
27646 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
27647 (V_lane_reg): Likewise.
27648 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
27650 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
27651 (vfmal_lane_low<mode>_intrinsic,
27652 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
27653 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
27654 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
27655 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
27656 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
27657 vfmsl_lane_high<mode>_intrinsic): New define_insns.
27659 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27661 * config/arm/arm-cpus.in (fp16fml): New feature.
27662 (ALL_SIMD): Add fp16fml.
27663 (armv8.2-a): Add fp16fml as an option.
27664 (armv8.3-a): Likewise.
27665 (armv8.4-a): Add fp16fml as part of fp16.
27666 * config/arm/arm.h (TARGET_FP16FML): Define.
27667 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
27669 * config/arm/arm-modes.def (V2HF): Define.
27670 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
27671 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
27672 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
27673 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
27674 vfmsl_low, vfmsl_high): New set of builtins.
27675 * config/arm/iterators.md (PLUSMINUS): New code iterator.
27676 (vfml_op): New code attribute.
27677 (VFMLHALVES): New int iterator.
27678 (VFML, VFMLSEL): New mode attributes.
27679 (V_reg): Define mapping for V2HF.
27680 (V_hi, V_lo): New mode attributes.
27681 (VF_constraint): Likewise.
27682 (vfml_half, vfml_half_selector): New int attributes.
27683 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
27685 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
27686 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
27688 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
27689 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
27690 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
27691 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
27693 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
27694 Document new effective target and option set.
27696 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27698 * config/arm/arm-cpus.in (armv8_4): New feature.
27699 (ARMv8_4a): New fgroup.
27700 (armv8.4-a): New arch.
27701 * config/arm/arm-tables.opt: Regenerate.
27702 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
27703 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
27704 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
27705 Add matching rules for -march=armv8.4-a and extensions.
27706 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
27708 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
27711 * config/rx/rx.md (BW): New mode attribute.
27712 (sync_lock_test_and_setsi): Add mode suffix to insn output.
27714 2018-01-11 Richard Biener <rguenther@suse.de>
27716 PR tree-optimization/83435
27717 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
27718 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
27719 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
27721 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27722 Alan Hayward <alan.hayward@arm.com>
27723 David Sherwood <david.sherwood@arm.com>
27725 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
27727 (aarch64_classify_address): Initialize it. Track polynomial offsets.
27728 (aarch64_print_address_internal): Use it to check for a zero offset.
27730 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27731 Alan Hayward <alan.hayward@arm.com>
27732 David Sherwood <david.sherwood@arm.com>
27734 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
27735 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
27736 Return a poly_int64 rather than a HOST_WIDE_INT.
27737 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
27738 rather than a HOST_WIDE_INT.
27739 * config/aarch64/aarch64.h (aarch64_frame): Protect with
27740 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
27741 hard_fp_offset, frame_size, initial_adjust, callee_offset and
27742 final_offset from HOST_WIDE_INT to poly_int64.
27743 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
27744 to_constant when getting the number of units in an Advanced SIMD
27746 (aarch64_builtin_vectorized_function): Check for a constant number
27748 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
27750 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
27751 attribute instead of GET_MODE_NUNITS.
27752 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
27753 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
27754 GET_MODE_SIZE for fixed-size registers.
27755 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
27756 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
27757 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
27758 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
27759 (aarch64_print_operand, aarch64_print_address_internal)
27760 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
27761 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
27762 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
27763 Handle polynomial GET_MODE_SIZE.
27764 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
27765 wider than SImode without modification.
27766 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
27767 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
27768 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
27769 passing and returning SVE modes.
27770 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
27771 rather than GEN_INT.
27772 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
27773 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
27774 (aarch64_allocate_and_probe_stack_space): Likewise.
27775 (aarch64_layout_frame): Cope with polynomial offsets.
27776 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
27777 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
27778 polynomial offsets.
27779 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
27780 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
27781 poly_int64 rather than a HOST_WIDE_INT.
27782 (aarch64_get_separate_components, aarch64_process_components)
27783 (aarch64_expand_prologue, aarch64_expand_epilogue)
27784 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
27785 (aarch64_anchor_offset): New function, split out from...
27786 (aarch64_legitimize_address): ...here.
27787 (aarch64_builtin_vectorization_cost): Handle polynomial
27788 TYPE_VECTOR_SUBPARTS.
27789 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
27791 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
27792 number of elements from the PARALLEL rather than the mode.
27793 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
27794 rather than GET_MODE_BITSIZE.
27795 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
27796 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
27797 (aarch64_expand_vec_perm_const_1): Handle polynomial
27798 d->perm.length () and d->perm elements.
27799 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
27800 Apply to_constant to d->perm elements.
27801 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
27802 polynomial CONST_VECTOR_NUNITS.
27803 (aarch64_move_pointer): Take amount as a poly_int64 rather
27805 (aarch64_progress_pointer): Avoid temporary variable.
27806 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
27807 the mode attribute instead of GET_MODE.
27809 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27810 Alan Hayward <alan.hayward@arm.com>
27811 David Sherwood <david.sherwood@arm.com>
27813 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
27814 x exists before using it.
27815 (aarch64_add_constant_internal): Rename to...
27816 (aarch64_add_offset_1): ...this. Replace regnum with separate
27817 src and dest rtxes. Handle the case in which they're different,
27818 including when the offset is zero. Replace scratchreg with an rtx.
27819 Use 2 additions if there is no spare register into which we can
27820 move a 16-bit constant.
27821 (aarch64_add_constant): Delete.
27822 (aarch64_add_offset): Replace reg with separate src and dest
27823 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
27824 Use aarch64_add_offset_1.
27825 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
27826 an rtx rather than an int. Take the delta as a poly_int64
27827 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
27828 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
27829 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
27830 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
27831 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
27832 and aarch64_add_sp.
27833 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
27834 aarch64_add_constant.
27836 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27838 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
27839 Use scalar_float_mode.
27841 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27843 * config/aarch64/aarch64-simd.md
27844 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
27845 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
27846 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
27847 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
27848 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
27849 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
27850 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
27851 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
27852 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
27853 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
27855 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27858 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
27859 targ_options->x_arm_arch_string is non NULL.
27861 2018-01-11 Tamar Christina <tamar.christina@arm.com>
27863 * config/aarch64/aarch64.h
27864 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
27866 2018-01-11 Sudakshina Das <sudi.das@arm.com>
27869 * expmed.c (emit_store_flag_force): Swap if const op0
27870 and change VOIDmode to mode of op0.
27872 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27874 PR rtl-optimization/83761
27875 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
27876 than bytes to mode_for_size.
27878 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
27880 PR middle-end/83189
27881 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
27882 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
27885 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
27887 PR middle-end/83575
27888 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
27889 when in layout mode.
27890 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
27891 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
27894 2018-01-10 Michael Collison <michael.collison@arm.com>
27896 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
27897 * config/aarch64/aarch64-option-extension.def: Add
27898 AARCH64_OPT_EXTENSION of 'fp16fml'.
27899 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27900 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
27901 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
27902 * config/aarch64/constraints.md (Ui7): New constraint.
27903 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
27904 (VFMLA_SEL_W): Ditto.
27907 (VFMLA16_LOW): New int iterator.
27908 (VFMLA16_HIGH): Ditto.
27909 (UNSPEC_FMLAL): New unspec.
27910 (UNSPEC_FMLSL): Ditto.
27911 (UNSPEC_FMLAL2): Ditto.
27912 (UNSPEC_FMLSL2): Ditto.
27913 (f16mac): New code attribute.
27914 * config/aarch64/aarch64-simd-builtins.def
27915 (aarch64_fmlal_lowv2sf): Ditto.
27916 (aarch64_fmlsl_lowv2sf): Ditto.
27917 (aarch64_fmlalq_lowv4sf): Ditto.
27918 (aarch64_fmlslq_lowv4sf): Ditto.
27919 (aarch64_fmlal_highv2sf): Ditto.
27920 (aarch64_fmlsl_highv2sf): Ditto.
27921 (aarch64_fmlalq_highv4sf): Ditto.
27922 (aarch64_fmlslq_highv4sf): Ditto.
27923 (aarch64_fmlal_lane_lowv2sf): Ditto.
27924 (aarch64_fmlsl_lane_lowv2sf): Ditto.
27925 (aarch64_fmlal_laneq_lowv2sf): Ditto.
27926 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
27927 (aarch64_fmlalq_lane_lowv4sf): Ditto.
27928 (aarch64_fmlsl_lane_lowv4sf): Ditto.
27929 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
27930 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
27931 (aarch64_fmlal_lane_highv2sf): Ditto.
27932 (aarch64_fmlsl_lane_highv2sf): Ditto.
27933 (aarch64_fmlal_laneq_highv2sf): Ditto.
27934 (aarch64_fmlsl_laneq_highv2sf): Ditto.
27935 (aarch64_fmlalq_lane_highv4sf): Ditto.
27936 (aarch64_fmlsl_lane_highv4sf): Ditto.
27937 (aarch64_fmlalq_laneq_highv4sf): Ditto.
27938 (aarch64_fmlsl_laneq_highv4sf): Ditto.
27939 * config/aarch64/aarch64-simd.md:
27940 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
27941 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
27942 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
27943 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
27944 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
27945 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
27946 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
27947 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
27948 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
27949 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
27950 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
27951 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
27952 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
27953 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
27954 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
27955 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
27956 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
27957 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
27958 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
27959 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
27960 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
27961 (vfmlsl_low_u32): Ditto.
27962 (vfmlalq_low_u32): Ditto.
27963 (vfmlslq_low_u32): Ditto.
27964 (vfmlal_high_u32): Ditto.
27965 (vfmlsl_high_u32): Ditto.
27966 (vfmlalq_high_u32): Ditto.
27967 (vfmlslq_high_u32): Ditto.
27968 (vfmlal_lane_low_u32): Ditto.
27969 (vfmlsl_lane_low_u32): Ditto.
27970 (vfmlal_laneq_low_u32): Ditto.
27971 (vfmlsl_laneq_low_u32): Ditto.
27972 (vfmlalq_lane_low_u32): Ditto.
27973 (vfmlslq_lane_low_u32): Ditto.
27974 (vfmlalq_laneq_low_u32): Ditto.
27975 (vfmlslq_laneq_low_u32): Ditto.
27976 (vfmlal_lane_high_u32): Ditto.
27977 (vfmlsl_lane_high_u32): Ditto.
27978 (vfmlal_laneq_high_u32): Ditto.
27979 (vfmlsl_laneq_high_u32): Ditto.
27980 (vfmlalq_lane_high_u32): Ditto.
27981 (vfmlslq_lane_high_u32): Ditto.
27982 (vfmlalq_laneq_high_u32): Ditto.
27983 (vfmlslq_laneq_high_u32): Ditto.
27984 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
27985 (AARCH64_FL_FOR_ARCH8_4): New.
27986 (AARCH64_ISA_F16FML): New ISA flag.
27987 (TARGET_F16FML): New feature flag for fp16fml.
27988 (doc/invoke.texi): Document new fp16fml option.
27990 2018-01-10 Michael Collison <michael.collison@arm.com>
27992 * config/aarch64/aarch64-builtins.c:
27993 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
27994 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27995 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
27996 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
27997 (AARCH64_ISA_SHA3): New ISA flag.
27998 (TARGET_SHA3): New feature flag for sha3.
27999 * config/aarch64/iterators.md (sha512_op): New int attribute.
28000 (CRYPTO_SHA512): New int iterator.
28001 (UNSPEC_SHA512H): New unspec.
28002 (UNSPEC_SHA512H2): Ditto.
28003 (UNSPEC_SHA512SU0): Ditto.
28004 (UNSPEC_SHA512SU1): Ditto.
28005 * config/aarch64/aarch64-simd-builtins.def
28006 (aarch64_crypto_sha512hqv2di): New builtin.
28007 (aarch64_crypto_sha512h2qv2di): Ditto.
28008 (aarch64_crypto_sha512su0qv2di): Ditto.
28009 (aarch64_crypto_sha512su1qv2di): Ditto.
28010 (aarch64_eor3qv8hi): Ditto.
28011 (aarch64_rax1qv2di): Ditto.
28012 (aarch64_xarqv2di): Ditto.
28013 (aarch64_bcaxqv8hi): Ditto.
28014 * config/aarch64/aarch64-simd.md:
28015 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
28016 (aarch64_crypto_sha512su0qv2di): Ditto.
28017 (aarch64_crypto_sha512su1qv2di): Ditto.
28018 (aarch64_eor3qv8hi): Ditto.
28019 (aarch64_rax1qv2di): Ditto.
28020 (aarch64_xarqv2di): Ditto.
28021 (aarch64_bcaxqv8hi): Ditto.
28022 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
28023 (vsha512h2q_u64): Ditto.
28024 (vsha512su0q_u64): Ditto.
28025 (vsha512su1q_u64): Ditto.
28026 (veor3q_u16): Ditto.
28027 (vrax1q_u64): Ditto.
28028 (vxarq_u64): Ditto.
28029 (vbcaxq_u16): Ditto.
28030 * config/arm/types.md (crypto_sha512): New type attribute.
28031 (crypto_sha3): Ditto.
28032 (doc/invoke.texi): Document new sha3 option.
28034 2018-01-10 Michael Collison <michael.collison@arm.com>
28036 * config/aarch64/aarch64-builtins.c:
28037 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
28038 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
28039 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
28040 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
28041 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
28042 (AARCH64_ISA_SM4): New ISA flag.
28043 (TARGET_SM4): New feature flag for sm4.
28044 * config/aarch64/aarch64-simd-builtins.def
28045 (aarch64_sm3ss1qv4si): Ditto.
28046 (aarch64_sm3tt1aq4si): Ditto.
28047 (aarch64_sm3tt1bq4si): Ditto.
28048 (aarch64_sm3tt2aq4si): Ditto.
28049 (aarch64_sm3tt2bq4si): Ditto.
28050 (aarch64_sm3partw1qv4si): Ditto.
28051 (aarch64_sm3partw2qv4si): Ditto.
28052 (aarch64_sm4eqv4si): Ditto.
28053 (aarch64_sm4ekeyqv4si): Ditto.
28054 * config/aarch64/aarch64-simd.md:
28055 (aarch64_sm3ss1qv4si): Ditto.
28056 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
28057 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
28058 (aarch64_sm4eqv4si): Ditto.
28059 (aarch64_sm4ekeyqv4si): Ditto.
28060 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
28061 (sm3part_op): Ditto.
28062 (CRYPTO_SM3TT): Ditto.
28063 (CRYPTO_SM3PART): Ditto.
28064 (UNSPEC_SM3SS1): New unspec.
28065 (UNSPEC_SM3TT1A): Ditto.
28066 (UNSPEC_SM3TT1B): Ditto.
28067 (UNSPEC_SM3TT2A): Ditto.
28068 (UNSPEC_SM3TT2B): Ditto.
28069 (UNSPEC_SM3PARTW1): Ditto.
28070 (UNSPEC_SM3PARTW2): Ditto.
28071 (UNSPEC_SM4E): Ditto.
28072 (UNSPEC_SM4EKEY): Ditto.
28073 * config/aarch64/constraints.md (Ui2): New constraint.
28074 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
28075 * config/arm/types.md (crypto_sm3): New type attribute.
28076 (crypto_sm4): Ditto.
28077 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
28078 (vsm3tt1aq_u32): Ditto.
28079 (vsm3tt1bq_u32): Ditto.
28080 (vsm3tt2aq_u32): Ditto.
28081 (vsm3tt2bq_u32): Ditto.
28082 (vsm3partw1q_u32): Ditto.
28083 (vsm3partw2q_u32): Ditto.
28084 (vsm4eq_u32): Ditto.
28085 (vsm4ekeyq_u32): Ditto.
28086 (doc/invoke.texi): Document new sm4 option.
28088 2018-01-10 Michael Collison <michael.collison@arm.com>
28090 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
28091 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
28092 (AARCH64_FL_FOR_ARCH8_4): New.
28093 (AARCH64_FL_V8_4): New flag.
28094 (doc/invoke.texi): Document new armv8.4-a option.
28096 2018-01-10 Michael Collison <michael.collison@arm.com>
28098 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
28099 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
28100 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
28101 * config/aarch64/aarch64-option-extension.def: Add
28102 AARCH64_OPT_EXTENSION of 'sha2'.
28103 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
28104 (crypto): Disable sha2 and aes if crypto disabled.
28105 (crypto): Enable aes and sha2 if enabled.
28106 (simd): Disable sha2 and aes if simd disabled.
28107 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
28109 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
28110 (TARGET_SHA2): New feature flag for sha2.
28111 (TARGET_AES): New feature flag for aes.
28112 * config/aarch64/aarch64-simd.md:
28113 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
28114 conditional on TARGET_AES.
28115 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
28116 (aarch64_crypto_sha1hsi): Make pattern conditional
28118 (aarch64_crypto_sha1hv4si): Ditto.
28119 (aarch64_be_crypto_sha1hv4si): Ditto.
28120 (aarch64_crypto_sha1su1v4si): Ditto.
28121 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
28122 (aarch64_crypto_sha1su0v4si): Ditto.
28123 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
28124 (aarch64_crypto_sha256su0v4si): Ditto.
28125 (aarch64_crypto_sha256su1v4si): Ditto.
28126 (doc/invoke.texi): Document new aes and sha2 options.
28128 2018-01-10 Martin Sebor <msebor@redhat.com>
28130 PR tree-optimization/83781
28131 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
28134 2018-01-11 Martin Sebor <msebor@gmail.com>
28135 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28137 PR tree-optimization/83501
28138 PR tree-optimization/81703
28140 * tree-ssa-strlen.c (get_string_cst): Rename...
28141 (get_string_len): ...to this. Handle global constants.
28142 (handle_char_store): Adjust.
28144 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
28145 Jim Wilson <jimw@sifive.com>
28147 * config/riscv/riscv-protos.h (riscv_output_return): New.
28148 * config/riscv/riscv.c (struct machine_function): New naked_p field.
28149 (riscv_attribute_table, riscv_output_return),
28150 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
28151 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
28152 (riscv_compute_frame_info): Only compute frame->mask if not a naked
28154 (riscv_expand_prologue): Add early return for naked function.
28155 (riscv_expand_epilogue): Likewise.
28156 (riscv_function_ok_for_sibcall): Return false for naked function.
28157 (riscv_set_current_function): New.
28158 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
28159 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
28160 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
28161 * doc/extend.texi (RISC-V Function Attributes): New.
28163 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
28165 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
28166 check for 128-bit long double before checking TCmode.
28167 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
28168 128-bit long doubles before checking TFmode or TCmode.
28169 (FLOAT128_IBM_P): Likewise.
28171 2018-01-10 Martin Sebor <msebor@redhat.com>
28173 PR tree-optimization/83671
28174 * builtins.c (c_strlen): Unconditionally return zero for the empty
28176 Use -Warray-bounds for warnings.
28177 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
28178 for non-constant array indices with COMPONENT_REF, arrays of
28179 arrays, and pointers to arrays.
28180 (gimple_fold_builtin_strlen): Determine and set length range for
28181 non-constant character arrays.
28183 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
28185 PR middle-end/81897
28186 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
28189 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
28191 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
28193 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
28196 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
28197 VECTOR_MEM_ALTIVEC_OR_VSX_P.
28198 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
28199 indexed_or_indirect_operand predicate.
28200 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
28201 (*vsx_le_perm_load_v8hi): Likewise.
28202 (*vsx_le_perm_load_v16qi): Likewise.
28203 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
28204 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
28205 (*vsx_le_perm_store_v8hi): Likewise.
28206 (*vsx_le_perm_store_v16qi): Likewise.
28207 (eight unnamed splitters): Likewise.
28209 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
28211 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
28212 * config/rs6000/emmintrin.h: Likewise.
28213 * config/rs6000/mmintrin.h: Likewise.
28214 * config/rs6000/xmmintrin.h: Likewise.
28216 2018-01-10 David Malcolm <dmalcolm@redhat.com>
28219 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
28221 * tree.c (tree_nop_conversion): Return true for location wrapper
28223 (maybe_wrap_with_location): New function.
28224 (selftest::check_strip_nops): New function.
28225 (selftest::test_location_wrappers): New function.
28226 (selftest::tree_c_tests): Call it.
28227 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
28228 (maybe_wrap_with_location): New decl.
28229 (EXPR_LOCATION_WRAPPER_P): New macro.
28230 (location_wrapper_p): New inline function.
28231 (tree_strip_any_location_wrapper): New inline function.
28233 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
28236 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
28237 stack_realign_offset for the largest alignment of stack slot
28239 (ix86_find_max_used_stack_alignment): New function.
28240 (ix86_finalize_stack_frame_flags): Use it. Set
28241 max_used_stack_alignment if we don't realign stack.
28242 * config/i386/i386.h (machine_function): Add
28243 max_used_stack_alignment.
28245 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
28247 * config/arm/arm.opt (-mbranch-cost): New option.
28248 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
28251 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
28254 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
28255 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
28257 2018-01-10 Richard Biener <rguenther@suse.de>
28260 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
28261 early out so it also covers the case where we have a non-NULL
28264 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
28266 PR tree-optimization/83753
28267 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
28268 for non-strided grouped accesses if the number of elements is 1.
28270 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
28273 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
28274 * i386.h (TARGET_USE_GATHER): Define.
28275 * x86-tune.def (X86_TUNE_USE_GATHER): New.
28277 2018-01-10 Martin Liska <mliska@suse.cz>
28280 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
28281 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
28283 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
28284 CLEANUP_NO_PARTITIONING is not set.
28286 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
28288 * doc/rtl.texi: Remove documentation of (const ...) wrappers
28289 for vectors, as a partial revert of r254296.
28290 * rtl.h (const_vec_p): Delete.
28291 (const_vec_duplicate_p): Don't test for vector CONSTs.
28292 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
28293 * expmed.c (make_tree): Likewise.
28296 * common.md (E, F): Use CONSTANT_P instead of checking for
28298 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
28299 checking for CONST_VECTOR.
28301 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
28303 PR middle-end/83575
28304 * predict.c (force_edge_cold): Handle in more sane way edges
28305 with no prediction.
28307 2018-01-09 Carl Love <cel@us.ibm.com>
28309 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
28311 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
28312 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
28313 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
28314 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
28315 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
28316 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
28317 * config/rs6000/rs6000-protos.h: Add extern defition for
28318 rs6000_generate_float2_double_code.
28319 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
28321 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
28322 (float2_v2df): Add define_expand.
28324 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
28327 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
28328 op_mode in the force_to_mode call.
28330 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
28332 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
28333 instead of checking each element individually.
28334 (aarch64_evpc_uzp): Likewise.
28335 (aarch64_evpc_zip): Likewise.
28336 (aarch64_evpc_ext): Likewise.
28337 (aarch64_evpc_rev): Likewise.
28338 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
28339 instead of checking each element individually. Return true without
28341 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
28342 whether all selected elements come from the same input, instead of
28343 checking each element individually. Remove calls to gen_rtx_REG,
28344 start_sequence and end_sequence and instead assert that no rtl is
28347 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
28349 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
28350 order of HIGH and CONST checks.
28352 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
28354 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
28355 if the destination isn't an SSA_NAME.
28357 2018-01-09 Richard Biener <rguenther@suse.de>
28359 PR tree-optimization/83668
28360 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
28362 (canonicalize_loop_form): ... here, renamed from ...
28363 (canonicalize_loop_closed_ssa_form): ... this and amended to
28364 swap successor edges for loop exit blocks to make us use
28365 the RPO order we need for initial schedule generation.
28367 2018-01-09 Joseph Myers <joseph@codesourcery.com>
28369 PR tree-optimization/64811
28370 * match.pd: When optimizing comparisons with Inf, avoid
28371 introducing or losing exceptions from comparisons with NaN.
28373 2018-01-09 Martin Liska <mliska@suse.cz>
28376 * asan.c (shadow_mem_size): Add gcc_assert.
28378 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
28380 Don't save registers in main().
28383 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
28384 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
28385 * config/avr/avr.c (avr_set_current_function): Don't error if
28386 naked, OS_task or OS_main are specified at the same time.
28387 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
28389 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
28391 * common/config/avr/avr-common.c (avr_option_optimization_table):
28392 Switch on -mmain-is-OS_task for optimizing compilations.
28394 2018-01-09 Richard Biener <rguenther@suse.de>
28396 PR tree-optimization/83572
28397 * graphite.c: Include cfganal.h.
28398 (graphite_transform_loops): Connect infinite loops to exit
28399 and remove fake edges at the end.
28401 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
28403 * ipa-inline.c (edge_badness): Revert accidental checkin.
28405 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
28408 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
28409 symbols; not inline clones.
28411 2018-01-09 Jakub Jelinek <jakub@redhat.com>
28414 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
28415 hard registers. Formatting fixes.
28417 PR preprocessor/83722
28418 * gcc.c (try_generate_repro): Pass
28419 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
28420 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
28423 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
28424 Kito Cheng <kito.cheng@gmail.com>
28426 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
28427 (riscv_leaf_function_p): Delete.
28428 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
28430 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28432 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
28434 (do_ifelse): New function.
28435 (do_isel): New function.
28436 (do_sub3): New function.
28437 (do_add3): New function.
28438 (do_load_mask_compare): New function.
28439 (do_overlap_load_compare): New function.
28440 (expand_compare_loop): New function.
28441 (expand_block_compare): Call expand_compare_loop() when appropriate.
28442 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
28443 option description.
28444 (-mblock-compare-inline-loop-limit): New option.
28446 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28449 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
28450 Reverse order of second and third operands in first alternative.
28451 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
28452 of first and second elements in UNSPEC_VPERMR vector.
28453 (altivec_expand_vec_perm_le): Likewise.
28455 2018-01-08 Jeff Law <law@redhat.com>
28457 PR rtl-optimizatin/81308
28458 * tree-switch-conversion.c (cfg_altered): New file scoped static.
28459 (process_switch): If group_case_labels makes a change, then set
28461 (pass_convert_switch::execute): If a switch is converted, then
28462 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
28464 PR rtl-optimization/81308
28465 * recog.c (split_all_insns): Conditionally cleanup the CFG after
28468 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
28470 PR target/83663 - Revert r255946
28471 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
28472 generation for cases where splatting a value is not useful.
28473 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
28474 across a vec_duplicate and a paradoxical subreg forming a vector
28475 mode to a vec_concat.
28477 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28479 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
28480 -march=armv8.3-a variants.
28481 * config/arm/t-multilib: Likewise.
28482 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
28484 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28486 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
28488 (cceq_ior_compare_complement): Give it a name so I can use it, and
28489 change boolean_or_operator predicate to boolean_operator so it can
28490 be used to generate a crand.
28491 (eqne): New code iterator.
28492 (bd/bd_neg): New code_attrs.
28493 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
28494 a single define_insn.
28495 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
28496 decrement (bdnzt/bdnzf/bdzt/bdzf).
28497 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
28498 with the new names of the branch decrement patterns, and added the
28499 names of the branch decrement conditional patterns.
28501 2018-01-08 Richard Biener <rguenther@suse.de>
28503 PR tree-optimization/83563
28504 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
28507 2018-01-08 Richard Biener <rguenther@suse.de>
28509 PR middle-end/83713
28510 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
28512 2018-01-08 Richard Biener <rguenther@suse.de>
28514 PR tree-optimization/83685
28515 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
28516 references to abnormals.
28518 2018-01-08 Richard Biener <rguenther@suse.de>
28521 * dwarf2out.c (output_indirect_strings): Handle empty
28522 skeleton_debug_str_hash.
28523 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
28525 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
28527 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
28528 (emit_store_direct): Likewise.
28529 (arc_trampoline_adjust_address): Likewise.
28530 (arc_asm_trampoline_template): New function.
28531 (arc_initialize_trampoline): Use asm_trampoline_template.
28532 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
28533 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
28534 * config/arc/arc.md (flush_icache): Delete pattern.
28536 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
28538 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
28539 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
28542 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
28545 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
28546 by not USED_FOR_TARGET.
28547 (make_pass_resolve_sw_modes): Likewise.
28549 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
28551 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
28554 2018-01-08 Richard Biener <rguenther@suse.de>
28556 PR middle-end/83580
28557 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
28559 2018-01-08 Richard Biener <rguenther@suse.de>
28561 PR middle-end/83517
28562 * match.pd ((t * 2) / 2) -> t): Add missing :c.
28564 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
28566 PR middle-end/81897
28567 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
28568 basic blocks with a small number of successors.
28569 (convert_control_dep_chain_into_preds): Improve handling of
28571 (dump_predicates): Split apart into...
28572 (dump_pred_chain): ...here...
28573 (dump_pred_info): ...and here.
28574 (can_one_predicate_be_invalidated_p): Add debugging printfs.
28575 (can_chain_union_be_invalidated_p): Improve check for invalidation
28577 (uninit_uses_cannot_happen): Avoid unnecessary if
28578 convert_control_dep_chain_into_preds yielded nothing.
28580 2018-01-06 Martin Sebor <msebor@redhat.com>
28582 PR tree-optimization/83640
28583 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
28584 subtracting negative offset from size.
28585 (builtin_access::overlap): Adjust offset bounds of the access to fall
28586 within the size of the object if possible.
28588 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
28590 PR rtl-optimization/83699
28591 * expmed.c (extract_bit_field_1): Restrict the vector usage of
28592 extract_bit_field_as_subreg to cases in which the extracted
28593 value is also a vector.
28595 * lra-constraints.c (process_alt_operands): Test for the equivalence
28596 substitutions when detecting a possible reload cycle.
28598 2018-01-06 Jakub Jelinek <jakub@redhat.com>
28601 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
28602 by default if flag_selective_schedling{,2}. Formatting fixes.
28604 PR rtl-optimization/83682
28605 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
28606 if it has non-VECTOR_MODE element mode.
28607 (vec_duplicate_p): Likewise.
28609 PR middle-end/83694
28610 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
28611 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
28613 2018-01-05 Jakub Jelinek <jakub@redhat.com>
28616 * config/i386/i386-builtin.def
28617 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
28618 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
28619 Require also OPTION_MASK_ISA_AVX512F in addition to
28620 OPTION_MASK_ISA_GFNI.
28621 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
28622 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
28623 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
28624 to OPTION_MASK_ISA_GFNI.
28625 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
28626 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
28627 OPTION_MASK_ISA_AVX512BW.
28628 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
28629 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
28630 addition to OPTION_MASK_ISA_GFNI.
28631 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
28632 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
28633 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
28634 to OPTION_MASK_ISA_GFNI.
28635 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
28636 a requirement for all ISAs rather than any of them with a few
28638 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
28640 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
28641 bitmasks to be enabled with 3 exceptions, instead of requiring any
28642 enabled ISA with lots of exceptions.
28643 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
28644 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
28645 Change avx512bw in isa attribute to avx512f.
28646 * config/i386/sgxintrin.h: Add license boilerplate.
28647 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
28648 to __AVX512F__ and __AVX512VL to __AVX512VL__.
28649 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
28650 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
28652 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
28653 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
28654 temporarily sse2 rather than sse if not enabled already.
28657 * config/i386/sse.md (VI248_VLBW): Rename to ...
28658 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
28659 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
28660 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
28661 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
28662 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
28663 mode iterator instead of VI248_VLBW.
28665 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
28667 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
28668 (record_modified): Skip clobbers; add debug output.
28669 (param_change_prob): Use sreal frequencies.
28671 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
28673 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
28674 punt for user-aligned variables.
28676 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
28678 * tree-chrec.c (chrec_contains_symbols): Return true for
28681 2018-01-05 Sudakshina Das <sudi.das@arm.com>
28684 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
28685 of (x|y) == x for BICS pattern.
28687 2018-01-05 Jakub Jelinek <jakub@redhat.com>
28689 PR tree-optimization/83605
28690 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
28691 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
28694 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
28696 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
28697 * config/epiphany/rtems.h: New file.
28699 2018-01-04 Jakub Jelinek <jakub@redhat.com>
28700 Uros Bizjak <ubizjak@gmail.com>
28703 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
28704 QIreg_operand instead of register_operand predicate.
28705 * config/i386/i386.c (ix86_rop_should_change_byte_p,
28706 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
28707 comments instead of -fmitigate[-_]rop.
28709 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28712 * cgraphunit.c (symbol_table::compile): Switch to text_section
28713 before calling assembly_start debug hook.
28714 * run-rtl-passes.c (run_rtl_passes): Likewise.
28717 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28719 * tree-vrp.c (extract_range_from_binary_expr_1): Check
28720 range_int_cst_p rather than !symbolic_range_p before calling
28721 extract_range_from_multiplicative_op_1.
28723 2018-01-04 Jeff Law <law@redhat.com>
28725 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
28726 redundant test in assertion.
28728 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28730 * doc/rtl.texi: Document machine_mode wrapper classes.
28732 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28734 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
28735 using tree_to_uhwi.
28737 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28739 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
28740 the VEC_PERM_EXPR fold to fail.
28742 2018-01-04 Jakub Jelinek <jakub@redhat.com>
28745 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
28746 to switched_sections.
28748 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28751 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
28752 test for d.testing.
28754 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
28757 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
28758 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
28760 2018-01-04 Jakub Jelinek <jakub@redhat.com>
28763 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
28764 is BLKmode and bitpos not zero or mode change is needed.
28766 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28769 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
28772 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
28775 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
28776 instead of MULT rtx. Update all corresponding splitters.
28777 (*saddl_se): Ditto.
28778 (*ssub<modesuffix>): Ditto.
28779 (*ssubl_se): Ditto.
28780 (*cmp_sadd_di): Update split patterns.
28781 (*cmp_sadd_si): Ditto.
28782 (*cmp_sadd_sidi): Ditto.
28783 (*cmp_ssub_di): Ditto.
28784 (*cmp_ssub_si): Ditto.
28785 (*cmp_ssub_sidi): Ditto.
28786 * config/alpha/predicates.md (const23_operand): New predicate.
28787 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
28788 Look for ASHIFT, not MULT inner operand.
28789 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
28791 2018-01-04 Martin Liska <mliska@suse.cz>
28793 PR gcov-profile/83669
28794 * gcov.c (output_intermediate_file): Add version to intermediate
28796 * doc/gcov.texi: Document new field 'version' in intermediate
28797 file format. Fix location of '-k' option of gcov command.
28799 2018-01-04 Martin Liska <mliska@suse.cz>
28802 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
28804 2018-01-04 Jakub Jelinek <jakub@redhat.com>
28806 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
28808 2018-01-03 Martin Sebor <msebor@redhat.com>
28810 PR tree-optimization/83655
28811 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
28812 checking calls with invalid arguments.
28814 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28816 * tree-vect-stmts.c (vect_get_store_rhs): New function.
28817 (vectorizable_mask_load_store): Delete.
28818 (vectorizable_call): Return false for masked loads and stores.
28819 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
28820 instead of gimple_assign_rhs1.
28821 (vectorizable_load): Handle IFN_MASK_LOAD.
28822 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
28824 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28826 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
28828 (vectorizable_mask_load_store): ...here.
28829 (vectorizable_load): ...and here.
28831 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28833 * tree-vect-stmts.c (vect_build_all_ones_mask)
28834 (vect_build_zero_merge_argument): New functions, split out from...
28835 (vectorizable_load): ...here.
28837 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28839 * tree-vect-stmts.c (vect_check_store_rhs): New function,
28841 (vectorizable_mask_load_store): ...here.
28842 (vectorizable_store): ...and here.
28844 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28846 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
28848 (vectorizable_mask_load_store): ...here.
28850 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28852 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
28853 (vect_model_store_cost): Take a vec_load_store_type instead of a
28855 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
28856 (vect_model_store_cost): Take a vec_load_store_type instead of a
28858 (vectorizable_mask_load_store): Update accordingly.
28859 (vectorizable_store): Likewise.
28860 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
28862 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28864 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
28865 IFN_MASK_LOAD calls here rather than...
28866 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
28868 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28869 Alan Hayward <alan.hayward@arm.com>
28870 David Sherwood <david.sherwood@arm.com>
28872 * expmed.c (extract_bit_field_1): For vector extracts,
28873 fall back to extract_bit_field_as_subreg if vec_extract
28876 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28877 Alan Hayward <alan.hayward@arm.com>
28878 David Sherwood <david.sherwood@arm.com>
28880 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
28881 they are variable or constant sized.
28882 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
28883 slots for constant-sized data.
28885 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28886 Alan Hayward <alan.hayward@arm.com>
28887 David Sherwood <david.sherwood@arm.com>
28889 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
28890 handling COND_EXPRs with boolean comparisons, try to find a better
28891 basis for the mask type than the boolean itself.
28893 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28895 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
28896 is calculated and how it can be overridden.
28897 * genmodes.c (max_bitsize_mode_any_mode): New variable.
28898 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
28900 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
28903 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28904 Alan Hayward <alan.hayward@arm.com>
28905 David Sherwood <david.sherwood@arm.com>
28907 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
28908 Remove the mode argument.
28909 (aarch64_simd_valid_immediate): Remove the mode and inverse
28911 * config/aarch64/iterators.md (bitsize): New iterator.
28912 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
28913 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
28914 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
28915 aarch64_simd_valid_immediate.
28916 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
28917 (aarch64_reg_or_bic_imm): Likewise.
28918 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
28919 with an insn_type enum and msl with a modifier_type enum.
28920 Replace element_width with a scalar_mode. Change the shift
28921 to unsigned int. Add constructors for scalar_float_mode and
28922 scalar_int_mode elements.
28923 (aarch64_vect_float_const_representable_p): Delete.
28924 (aarch64_can_const_movi_rtx_p)
28925 (aarch64_simd_scalar_immediate_valid_for_move)
28926 (aarch64_simd_make_constant): Update call to
28927 aarch64_simd_valid_immediate.
28928 (aarch64_advsimd_valid_immediate_hs): New function.
28929 (aarch64_advsimd_valid_immediate): Likewise.
28930 (aarch64_simd_valid_immediate): Remove mode and inverse
28931 arguments. Rewrite to use the above. Use const_vec_duplicate_p
28932 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
28933 and aarch64_float_const_representable_p on the result.
28934 (aarch64_output_simd_mov_immediate): Remove mode argument.
28935 Update call to aarch64_simd_valid_immediate and use of
28936 simd_immediate_info.
28937 (aarch64_output_scalar_simd_mov_immediate): Update call
28940 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28941 Alan Hayward <alan.hayward@arm.com>
28942 David Sherwood <david.sherwood@arm.com>
28944 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
28945 (mode_nunits): Likewise CONST_MODE_NUNITS.
28946 * machmode.def (ADJUST_NUNITS): Document.
28947 * genmodes.c (mode_data::need_nunits_adj): New field.
28948 (blank_mode): Update accordingly.
28949 (adj_nunits): New variable.
28950 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
28952 (emit_mode_size_inline): Set need_bytesize_adj for all modes
28953 listed in adj_nunits.
28954 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
28955 listed in adj_nunits. Don't emit case statements for such modes.
28956 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
28957 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
28958 nothing if adj_nunits is nonnull.
28959 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
28960 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
28961 (emit_mode_fbit): Update use of print_maybe_const_decl.
28962 (emit_move_size): Likewise. Treat the array as non-const
28964 (emit_mode_adjustments): Handle adj_nunits.
28966 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28968 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
28969 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
28970 (VECTOR_MODES): Use it.
28971 (make_vector_modes): Take the prefix as an argument.
28973 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28974 Alan Hayward <alan.hayward@arm.com>
28975 David Sherwood <david.sherwood@arm.com>
28977 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
28978 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
28979 for MODE_VECTOR_BOOL.
28980 * machmode.def (VECTOR_BOOL_MODE): Document.
28981 * genmodes.c (VECTOR_BOOL_MODE): New macro.
28982 (make_vector_bool_mode): New function.
28983 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
28985 * lto-streamer-in.c (lto_input_mode_table): Likewise.
28986 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
28988 * stor-layout.c (int_mode_for_mode): Likewise.
28989 * tree.c (build_vector_type_for_mode): Likewise.
28990 * varasm.c (output_constant_pool_2): Likewise.
28991 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
28992 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
28993 for MODE_VECTOR_BOOL.
28994 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
28995 of mode class checks.
28996 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
28997 instead of a list of mode class checks.
28998 (expand_vector_scalar_condition): Likewise.
28999 (type_for_widest_vector_mode): Handle BImode as an inner mode.
29001 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29002 Alan Hayward <alan.hayward@arm.com>
29003 David Sherwood <david.sherwood@arm.com>
29005 * machmode.h (mode_size): Change from unsigned short to
29007 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
29008 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
29009 or if measurement_type is not polynomial.
29010 (fixed_size_mode::includes_p): Check for constant-sized modes.
29011 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
29012 return a poly_uint16 rather than an unsigned short.
29013 (emit_mode_size): Change the type of mode_size from unsigned short
29014 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
29015 (emit_mode_adjustments): Cope with polynomial vector sizes.
29016 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
29018 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
29020 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
29021 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
29022 * caller-save.c (setup_save_areas): Likewise.
29023 (replace_reg_with_saved_mem): Likewise.
29024 * calls.c (emit_library_call_value_1): Likewise.
29025 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
29026 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
29027 (gen_lowpart_for_combine): Likewise.
29028 * convert.c (convert_to_integer_1): Likewise.
29029 * cse.c (equiv_constant, cse_insn): Likewise.
29030 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
29031 (cselib_subst_to_values): Likewise.
29032 * dce.c (word_dce_process_block): Likewise.
29033 * df-problems.c (df_word_lr_mark_ref): Likewise.
29034 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
29035 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
29036 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
29037 (rtl_for_decl_location): Likewise.
29038 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
29039 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
29040 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
29041 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
29042 (expand_expr_real_1): Likewise.
29043 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
29044 (pad_below): Likewise.
29045 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
29046 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
29047 * ira.c (get_subreg_tracking_sizes): Likewise.
29048 * ira-build.c (ira_create_allocno_objects): Likewise.
29049 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
29050 (ira_sort_regnos_for_alter_reg): Likewise.
29051 * ira-costs.c (record_operand_costs): Likewise.
29052 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
29053 (resolve_simple_move): Likewise.
29054 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
29055 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
29056 (lra_constraints): Likewise.
29057 (CONST_POOL_OK_P): Reject variable-sized modes.
29058 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
29059 (add_pseudo_to_slot, lra_spill): Likewise.
29060 * omp-low.c (omp_clause_aligned_alignment): Likewise.
29061 * optabs-query.c (get_best_extraction_insn): Likewise.
29062 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
29063 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
29064 (expand_mult_highpart, valid_multiword_target_p): Likewise.
29065 * recog.c (offsettable_address_addr_space_p): Likewise.
29066 * regcprop.c (maybe_mode_change): Likewise.
29067 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
29068 * regrename.c (build_def_use): Likewise.
29069 * regstat.c (dump_reg_info): Likewise.
29070 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
29071 (find_reloads, find_reloads_subreg_address): Likewise.
29072 * reload1.c (eliminate_regs_1): Likewise.
29073 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
29074 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
29075 (simplify_binary_operation_1, simplify_subreg): Likewise.
29076 * targhooks.c (default_function_arg_padding): Likewise.
29077 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
29078 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
29079 (verify_gimple_assign_ternary): Likewise.
29080 * tree-inline.c (estimate_move_cost): Likewise.
29081 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
29082 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
29083 (get_address_cost_ainc): Likewise.
29084 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
29085 (vect_supportable_dr_alignment): Likewise.
29086 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
29087 (vectorizable_reduction): Likewise.
29088 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
29089 (vectorizable_operation, vectorizable_load): Likewise.
29090 * tree.c (build_same_sized_truth_vector_type): Likewise.
29091 * valtrack.c (cleanup_auto_inc_dec): Likewise.
29092 * var-tracking.c (emit_note_insn_var_location): Likewise.
29093 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
29094 (ADDR_VEC_ALIGN): Likewise.
29096 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29097 Alan Hayward <alan.hayward@arm.com>
29098 David Sherwood <david.sherwood@arm.com>
29100 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
29102 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
29103 or if measurement_type is polynomial.
29104 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
29105 * combine.c (make_extraction): Likewise.
29106 * dse.c (find_shift_sequence): Likewise.
29107 * dwarf2out.c (mem_loc_descriptor): Likewise.
29108 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
29109 (extract_bit_field, extract_low_bits): Likewise.
29110 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
29111 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
29112 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
29113 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
29114 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
29115 * reload.c (find_reloads): Likewise.
29116 * reload1.c (alter_reg): Likewise.
29117 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
29118 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
29119 * tree-if-conv.c (predicate_mem_writes): Likewise.
29120 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
29121 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
29122 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
29123 * valtrack.c (dead_debug_insert_temp): Likewise.
29124 * varasm.c (mergeable_constant_section): Likewise.
29125 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
29127 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29128 Alan Hayward <alan.hayward@arm.com>
29129 David Sherwood <david.sherwood@arm.com>
29131 * expr.c (expand_assignment): Cope with polynomial mode sizes
29132 when assigning to a CONCAT.
29134 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29135 Alan Hayward <alan.hayward@arm.com>
29136 David Sherwood <david.sherwood@arm.com>
29138 * machmode.h (mode_precision): Change from unsigned short to
29140 (mode_to_precision): Return a poly_uint16 rather than an unsigned
29142 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
29143 or if measurement_type is not polynomial.
29144 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
29145 in which the mode is already known to be a scalar_int_mode.
29146 * genmodes.c (emit_mode_precision): Change the type of mode_precision
29147 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
29149 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
29150 for GET_MODE_PRECISION.
29151 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
29152 for GET_MODE_PRECISION.
29153 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
29155 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
29156 (expand_field_assignment, make_extraction): Likewise.
29157 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
29158 (get_last_value): Likewise.
29159 * convert.c (convert_to_integer_1): Likewise.
29160 * cse.c (cse_insn): Likewise.
29161 * expr.c (expand_expr_real_1): Likewise.
29162 * lra-constraints.c (simplify_operand_subreg): Likewise.
29163 * optabs-query.c (can_atomic_load_p): Likewise.
29164 * optabs.c (expand_atomic_load): Likewise.
29165 (expand_atomic_store): Likewise.
29166 * ree.c (combine_reaching_defs): Likewise.
29167 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
29168 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
29169 * tree.h (type_has_mode_precision_p): Likewise.
29170 * ubsan.c (instrument_si_overflow): Likewise.
29172 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29173 Alan Hayward <alan.hayward@arm.com>
29174 David Sherwood <david.sherwood@arm.com>
29176 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
29177 polynomial numbers of units.
29178 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
29179 (valid_vector_subparts_p): New function.
29180 (build_vector_type): Remove temporary shim and take the number
29181 of units as a poly_uint64 rather than an int.
29182 (build_opaque_vector_type): Take the number of units as a
29183 poly_uint64 rather than an int.
29184 * tree.c (build_vector_from_ctor): Handle polynomial
29185 TYPE_VECTOR_SUBPARTS.
29186 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
29187 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
29188 (build_vector_from_val): If the number of units is variable,
29189 use build_vec_duplicate_cst for constant operands and
29190 VEC_DUPLICATE_EXPR otherwise.
29191 (make_vector_type): Remove temporary is_constant ().
29192 (build_vector_type, build_opaque_vector_type): Take the number of
29193 units as a poly_uint64 rather than an int.
29194 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
29196 * cfgexpand.c (expand_debug_expr): Likewise.
29197 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
29198 (store_constructor, expand_expr_real_1): Likewise.
29199 (const_scalar_mask_from_tree): Likewise.
29200 * fold-const-call.c (fold_const_reduction): Likewise.
29201 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
29202 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
29203 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
29204 (fold_relational_const): Likewise.
29205 (native_interpret_vector): Likewise. Change the size from an
29206 int to an unsigned int.
29207 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
29208 TYPE_VECTOR_SUBPARTS.
29209 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
29210 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
29211 duplicating a non-constant operand into a variable-length vector.
29212 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
29213 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
29214 * ipa-icf.c (sem_variable::equals): Likewise.
29215 * match.pd: Likewise.
29216 * omp-simd-clone.c (simd_clone_subparts): Likewise.
29217 * print-tree.c (print_node): Likewise.
29218 * stor-layout.c (layout_type): Likewise.
29219 * targhooks.c (default_builtin_vectorization_cost): Likewise.
29220 * tree-cfg.c (verify_gimple_comparison): Likewise.
29221 (verify_gimple_assign_binary): Likewise.
29222 (verify_gimple_assign_ternary): Likewise.
29223 (verify_gimple_assign_single): Likewise.
29224 * tree-pretty-print.c (dump_generic_node): Likewise.
29225 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
29226 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
29227 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
29228 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
29229 (vect_shift_permute_load_chain): Likewise.
29230 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
29231 (expand_vector_condition, optimize_vector_constructor): Likewise.
29232 (lower_vec_perm, get_compute_type): Likewise.
29233 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
29234 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
29235 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
29236 (vect_recog_mask_conversion_pattern): Likewise.
29237 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
29238 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
29239 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
29240 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
29241 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
29242 (vectorizable_shift, vectorizable_operation, vectorizable_store)
29243 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
29244 (supportable_widening_operation): Likewise.
29245 (supportable_narrowing_operation): Likewise.
29246 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
29248 * varasm.c (output_constant): Likewise.
29250 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29251 Alan Hayward <alan.hayward@arm.com>
29252 David Sherwood <david.sherwood@arm.com>
29254 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
29255 so that both the length == 3 and length != 3 cases set up their
29256 own permute vectors. Add comments explaining why we know the
29257 number of elements is constant.
29258 (vect_permute_load_chain): Likewise.
29260 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29261 Alan Hayward <alan.hayward@arm.com>
29262 David Sherwood <david.sherwood@arm.com>
29264 * machmode.h (mode_nunits): Change from unsigned char to
29266 (ONLY_FIXED_SIZE_MODES): New macro.
29267 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
29268 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
29269 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
29271 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
29272 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
29273 or if measurement_type is not polynomial.
29274 * genmodes.c (ZERO_COEFFS): New macro.
29275 (emit_mode_nunits_inline): Make mode_nunits_inline return a
29277 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
29278 Use ZERO_COEFFS when emitting initializers.
29279 * data-streamer.h (bp_pack_poly_value): New function.
29280 (bp_unpack_poly_value): Likewise.
29281 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
29282 for GET_MODE_NUNITS.
29283 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
29284 for GET_MODE_NUNITS.
29285 * tree.c (make_vector_type): Remove temporary shim and make
29286 the real function take the number of units as a poly_uint64
29287 rather than an int.
29288 (build_vector_type_for_mode): Handle polynomial nunits.
29289 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
29290 * emit-rtl.c (const_vec_series_p_1): Likewise.
29291 (gen_rtx_CONST_VECTOR): Likewise.
29292 * fold-const.c (test_vec_duplicate_folding): Likewise.
29293 * genrecog.c (validate_pattern): Likewise.
29294 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
29295 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
29296 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
29297 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
29298 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
29299 * rtlanal.c (subreg_get_info): Likewise.
29300 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
29301 (vect_grouped_load_supported): Likewise.
29302 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
29303 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
29304 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
29305 (simplify_const_unary_operation, simplify_binary_operation_1)
29306 (simplify_const_binary_operation, simplify_ternary_operation)
29307 (test_vector_ops_duplicate, test_vector_ops): Likewise.
29308 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
29309 instead of CONST_VECTOR_NUNITS.
29310 * varasm.c (output_constant_pool_2): Likewise.
29311 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
29312 explicit-encoded elements in the XVEC for variable-length vectors.
29314 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29316 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
29318 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29319 Alan Hayward <alan.hayward@arm.com>
29320 David Sherwood <david.sherwood@arm.com>
29322 * coretypes.h (fixed_size_mode): Declare.
29323 (fixed_size_mode_pod): New typedef.
29324 * builtins.h (target_builtins::x_apply_args_mode)
29325 (target_builtins::x_apply_result_mode): Change type to
29326 fixed_size_mode_pod.
29327 * builtins.c (apply_args_size, apply_result_size, result_vector)
29328 (expand_builtin_apply_args_1, expand_builtin_apply)
29329 (expand_builtin_return): Update accordingly.
29331 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29333 * cse.c (hash_rtx_cb): Hash only the encoded elements.
29334 * cselib.c (cselib_hash_rtx): Likewise.
29335 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
29336 CONST_VECTOR encoding.
29338 2018-01-03 Jakub Jelinek <jakub@redhat.com>
29339 Jeff Law <law@redhat.com>
29342 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
29343 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
29344 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
29345 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
29348 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
29349 explicitly probe *sp in a noreturn function if there were any callee
29350 register saves or frame pointer is needed.
29352 2018-01-03 Jakub Jelinek <jakub@redhat.com>
29355 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
29356 BLKmode for ternary, binary or unary expressions.
29359 * var-tracking.c (delete_vta_debug_insn): New inline function.
29360 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
29361 insns from get_insns () to NULL instead of each bb separately.
29362 Use delete_vta_debug_insn. No longer static.
29363 (vt_debug_insns_local, variable_tracking_main_1): Adjust
29364 delete_vta_debug_insns callers.
29365 * rtl.h (delete_vta_debug_insns): Declare.
29366 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
29367 instead of variable_tracking_main.
29369 2018-01-03 Martin Sebor <msebor@redhat.com>
29371 PR tree-optimization/83603
29372 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
29373 arguments past the endof the argument list in functions declared
29374 without a prototype.
29375 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
29376 Avoid checking when arguments are null.
29378 2018-01-03 Martin Sebor <msebor@redhat.com>
29381 * doc/extend.texi (attribute const): Fix a typo.
29382 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
29383 issuing -Wsuggest-attribute for void functions.
29385 2018-01-03 Martin Sebor <msebor@redhat.com>
29387 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
29388 offset_int::from instead of wide_int::to_shwi.
29389 (maybe_diag_overlap): Remove assertion.
29390 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
29391 * gimple-ssa-sprintf.c (format_directive): Same.
29392 (parse_directive): Same.
29393 (sprintf_dom_walker::compute_format_length): Same.
29394 (try_substitute_return_value): Same.
29396 2018-01-03 Jeff Law <law@redhat.com>
29398 PR middle-end/83654
29399 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
29400 non-constant residual for zero at runtime and avoid probing in
29401 that case. Reorganize code for trailing problem to mirror handling
29404 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29406 PR tree-optimization/83501
29407 * tree-ssa-strlen.c (get_string_cst): New.
29408 (handle_char_store): Call get_string_cst.
29410 2018-01-03 Martin Liska <mliska@suse.cz>
29412 PR tree-optimization/83593
29413 * tree-ssa-strlen.c: Include tree-cfg.h.
29414 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
29415 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
29416 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
29418 (strlen_dom_walker::before_dom_children): Call
29419 gimple_purge_dead_eh_edges. Dump tranformation with details
29421 (strlen_dom_walker::before_dom_children): Update call by adding
29422 new argument cleanup_eh.
29423 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
29425 2018-01-03 Martin Liska <mliska@suse.cz>
29428 * cif-code.def (VARIADIC_THUNK): New enum value.
29429 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
29432 2018-01-03 Jan Beulich <jbeulich@suse.com>
29434 * sse.md (mov<mode>_internal): Tighten condition for when to use
29435 vmovdqu<ssescalarsize> for TI and OI modes.
29437 2018-01-03 Jakub Jelinek <jakub@redhat.com>
29439 Update copyright years.
29441 2018-01-03 Martin Liska <mliska@suse.cz>
29444 * ipa-visibility.c (function_and_variable_visibility): Skip
29445 functions with noipa attribure.
29447 2018-01-03 Jakub Jelinek <jakub@redhat.com>
29449 * gcc.c (process_command): Update copyright notice dates.
29450 * gcov-dump.c (print_version): Ditto.
29451 * gcov.c (print_version): Ditto.
29452 * gcov-tool.c (print_version): Ditto.
29453 * gengtype.c (create_file): Ditto.
29454 * doc/cpp.texi: Bump @copying's copyright year.
29455 * doc/cppinternals.texi: Ditto.
29456 * doc/gcc.texi: Ditto.
29457 * doc/gccint.texi: Ditto.
29458 * doc/gcov.texi: Ditto.
29459 * doc/install.texi: Ditto.
29460 * doc/invoke.texi: Ditto.
29462 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29464 * vector-builder.h (vector_builder::m_full_nelts): Change from
29465 unsigned int to poly_uint64.
29466 (vector_builder::full_nelts): Update prototype accordingly.
29467 (vector_builder::new_vector): Likewise.
29468 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
29469 (vector_builder::operator ==): Likewise.
29470 (vector_builder::finalize): Likewise.
29471 * int-vector-builder.h (int_vector_builder::int_vector_builder):
29472 Take the number of elements as a poly_uint64 rather than an
29474 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
29475 from unsigned int to poly_uint64.
29476 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
29477 (vec_perm_indices::new_vector): Likewise.
29478 (vec_perm_indices::length): Likewise.
29479 (vec_perm_indices::nelts_per_input): Likewise.
29480 (vec_perm_indices::input_nelts): Likewise.
29481 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
29482 number of elements per input as a poly_uint64 rather than an
29483 unsigned int. Use the original encoding for variable-length
29484 vectors, rather than clamping each individual element.
29485 For the second and subsequent elements in each pattern,
29486 clamp the step and base before clamping their sum.
29487 (vec_perm_indices::series_p): Handle polynomial element counts.
29488 (vec_perm_indices::all_in_range_p): Likewise.
29489 (vec_perm_indices_to_tree): Likewise.
29490 (vec_perm_indices_to_rtx): Likewise.
29491 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
29492 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
29493 (tree_vector_builder::new_binary_operation): Handle polynomial
29494 element counts. Return false if we need to know the number
29495 of elements at compile time.
29496 * fold-const.c (fold_vec_perm): Punt if the number of elements
29497 isn't known at compile time.
29499 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29501 * vec-perm-indices.h (vec_perm_builder): Change element type
29502 from HOST_WIDE_INT to poly_int64.
29503 (vec_perm_indices::element_type): Update accordingly.
29504 (vec_perm_indices::clamp): Handle polynomial element_types.
29505 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
29506 (vec_perm_indices::all_in_range_p): Likewise.
29507 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
29509 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
29510 polynomial vec_perm_indices element types.
29511 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
29512 * fold-const.c (fold_vec_perm): Likewise.
29513 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
29514 * tree-vect-generic.c (lower_vec_perm): Likewise.
29515 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
29516 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
29517 element type to HOST_WIDE_INT.
29519 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29520 Alan Hayward <alan.hayward@arm.com>
29521 David Sherwood <david.sherwood@arm.com>
29523 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
29524 rather than an int. Use plus_constant.
29525 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
29526 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
29528 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29529 Alan Hayward <alan.hayward@arm.com>
29530 David Sherwood <david.sherwood@arm.com>
29532 * calls.c (emit_call_1, expand_call): Change struct_value_size from
29533 a HOST_WIDE_INT to a poly_int64.
29535 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29536 Alan Hayward <alan.hayward@arm.com>
29537 David Sherwood <david.sherwood@arm.com>
29539 * calls.c (load_register_parameters): Cope with polynomial
29540 mode sizes. Require a constant size for BLKmode parameters
29541 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
29542 forces a parameter to be padded at the lsb end in order to
29543 fill a complete number of words, require the parameter size
29544 to be ordered wrt UNITS_PER_WORD.
29546 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29547 Alan Hayward <alan.hayward@arm.com>
29548 David Sherwood <david.sherwood@arm.com>
29550 * reload1.c (spill_stack_slot_width): Change element type
29551 from unsigned int to poly_uint64_pod.
29552 (alter_reg): Treat mode sizes as polynomial.
29554 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29555 Alan Hayward <alan.hayward@arm.com>
29556 David Sherwood <david.sherwood@arm.com>
29558 * reload.c (complex_word_subreg_p): New function.
29559 (reload_inner_reg_of_subreg, push_reload): Use it.
29561 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29562 Alan Hayward <alan.hayward@arm.com>
29563 David Sherwood <david.sherwood@arm.com>
29565 * lra-constraints.c (process_alt_operands): Reject matched
29566 operands whose sizes aren't ordered.
29567 (match_reload): Refer to this check here.
29569 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29570 Alan Hayward <alan.hayward@arm.com>
29571 David Sherwood <david.sherwood@arm.com>
29573 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
29574 that the mode size is in the set {1, 2, 4, 8, 16}.
29576 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29577 Alan Hayward <alan.hayward@arm.com>
29578 David Sherwood <david.sherwood@arm.com>
29580 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
29581 Use plus_constant instead of gen_rtx_PLUS.
29583 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29584 Alan Hayward <alan.hayward@arm.com>
29585 David Sherwood <david.sherwood@arm.com>
29587 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
29588 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
29589 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
29590 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
29591 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
29592 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
29593 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
29594 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
29595 * config/i386/i386.c (ix86_push_rounding): ...this new function.
29596 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
29598 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
29599 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
29600 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
29601 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
29602 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
29603 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
29604 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
29605 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
29606 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
29607 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
29609 * expr.c (emit_move_resolve_push): Treat the input and result
29610 of PUSH_ROUNDING as a poly_int64.
29611 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
29612 (emit_push_insn): Likewise.
29613 * lra-eliminations.c (mark_not_eliminable): Likewise.
29614 * recog.c (push_operand): Likewise.
29615 * reload1.c (elimination_effects): Likewise.
29616 * rtlanal.c (nonzero_bits1): Likewise.
29617 * calls.c (store_one_arg): Likewise. Require the padding to be
29618 known at compile time.
29620 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29621 Alan Hayward <alan.hayward@arm.com>
29622 David Sherwood <david.sherwood@arm.com>
29624 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
29625 Use plus_constant instead of gen_rtx_PLUS.
29627 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29628 Alan Hayward <alan.hayward@arm.com>
29629 David Sherwood <david.sherwood@arm.com>
29631 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
29632 rather than an int.
29634 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29635 Alan Hayward <alan.hayward@arm.com>
29636 David Sherwood <david.sherwood@arm.com>
29638 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
29639 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
29640 via stack temporaries. Treat the mode size as polynomial too.
29642 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29643 Alan Hayward <alan.hayward@arm.com>
29644 David Sherwood <david.sherwood@arm.com>
29646 * expr.c (expand_expr_real_2): When handling conversions involving
29647 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
29648 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
29649 as a poly_uint64 too.
29651 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29652 Alan Hayward <alan.hayward@arm.com>
29653 David Sherwood <david.sherwood@arm.com>
29655 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
29657 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29658 Alan Hayward <alan.hayward@arm.com>
29659 David Sherwood <david.sherwood@arm.com>
29661 * combine.c (can_change_dest_mode): Handle polynomial
29662 REGMODE_NATURAL_SIZE.
29663 * expmed.c (store_bit_field_1): Likewise.
29664 * expr.c (store_constructor): Likewise.
29665 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
29666 and polynomial REGMODE_NATURAL_SIZE.
29667 (gen_lowpart_common): Likewise.
29668 * reginfo.c (record_subregs_of_mode): Likewise.
29669 * rtlanal.c (read_modify_subreg_p): Likewise.
29671 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29672 Alan Hayward <alan.hayward@arm.com>
29673 David Sherwood <david.sherwood@arm.com>
29675 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
29676 numbers of elements.
29678 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29679 Alan Hayward <alan.hayward@arm.com>
29680 David Sherwood <david.sherwood@arm.com>
29682 * match.pd: Cope with polynomial numbers of vector elements.
29684 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29685 Alan Hayward <alan.hayward@arm.com>
29686 David Sherwood <david.sherwood@arm.com>
29688 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
29689 in a POINTER_PLUS_EXPR.
29691 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29692 Alan Hayward <alan.hayward@arm.com>
29693 David Sherwood <david.sherwood@arm.com>
29695 * omp-simd-clone.c (simd_clone_subparts): New function.
29696 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
29697 (ipa_simd_modify_function_body): Likewise.
29699 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29700 Alan Hayward <alan.hayward@arm.com>
29701 David Sherwood <david.sherwood@arm.com>
29703 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
29704 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
29705 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
29706 (expand_vector_condition, vector_element): Likewise.
29707 (subparts_gt): New function.
29708 (get_compute_type): Use subparts_gt.
29709 (count_type_subparts): Delete.
29710 (expand_vector_operations_1): Use subparts_gt instead of
29711 count_type_subparts.
29713 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29714 Alan Hayward <alan.hayward@arm.com>
29715 David Sherwood <david.sherwood@arm.com>
29717 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
29718 (vect_compile_time_alias): ...this new function. Do the calculation
29719 on poly_ints rather than trees.
29720 (vect_prune_runtime_alias_test_list): Update call accordingly.
29722 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29723 Alan Hayward <alan.hayward@arm.com>
29724 David Sherwood <david.sherwood@arm.com>
29726 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
29728 (vect_schedule_slp_instance): Likewise.
29730 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29731 Alan Hayward <alan.hayward@arm.com>
29732 David Sherwood <david.sherwood@arm.com>
29734 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
29735 constant and extern definitions for variable-length vectors.
29736 (vect_get_constant_vectors): Note that the number of units
29737 is known to be constant.
29739 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29740 Alan Hayward <alan.hayward@arm.com>
29741 David Sherwood <david.sherwood@arm.com>
29743 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
29744 of units as polynomial. Choose between WIDE and NARROW based
29747 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29748 Alan Hayward <alan.hayward@arm.com>
29749 David Sherwood <david.sherwood@arm.com>
29751 * tree-vect-stmts.c (simd_clone_subparts): New function.
29752 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
29754 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29755 Alan Hayward <alan.hayward@arm.com>
29756 David Sherwood <david.sherwood@arm.com>
29758 * tree-vect-stmts.c (vectorizable_call): Treat the number of
29759 vectors as polynomial. Use build_index_vector for
29760 IFN_GOMP_SIMD_LANE.
29762 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29763 Alan Hayward <alan.hayward@arm.com>
29764 David Sherwood <david.sherwood@arm.com>
29766 * tree-vect-stmts.c (get_load_store_type): Treat the number of
29767 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
29768 for variable-length vectors.
29769 (vectorizable_mask_load_store): Treat the number of units as
29770 polynomial, asserting that it is constant if the condition has
29771 already been enforced.
29772 (vectorizable_store, vectorizable_load): Likewise.
29774 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29775 Alan Hayward <alan.hayward@arm.com>
29776 David Sherwood <david.sherwood@arm.com>
29778 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
29779 of units as polynomial. Punt if we can't tell at compile time
29780 which vector contains the final result.
29782 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29783 Alan Hayward <alan.hayward@arm.com>
29784 David Sherwood <david.sherwood@arm.com>
29786 * tree-vect-loop.c (vectorizable_induction): Treat the number
29787 of units as polynomial. Punt on SLP inductions. Use an integer
29788 VEC_SERIES_EXPR for variable-length integer reductions. Use a
29789 cast of such a series for variable-length floating-point
29792 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29793 Alan Hayward <alan.hayward@arm.com>
29794 David Sherwood <david.sherwood@arm.com>
29796 * tree.h (build_index_vector): Declare.
29797 * tree.c (build_index_vector): New function.
29798 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
29799 of units as polynomial, forcibly converting it to a constant if
29800 vectorizable_reduction has already enforced the condition.
29801 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
29802 to create a {1,2,3,...} vector.
29803 (vectorizable_reduction): Treat the number of units as polynomial.
29804 Choose vectype_in based on the largest scalar element size rather
29805 than the smallest number of units. Enforce the restrictions
29808 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29809 Alan Hayward <alan.hayward@arm.com>
29810 David Sherwood <david.sherwood@arm.com>
29812 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
29813 number of units as polynomial.
29815 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29816 Alan Hayward <alan.hayward@arm.com>
29817 David Sherwood <david.sherwood@arm.com>
29819 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
29820 * target.def (autovectorize_vector_sizes): Return the vector sizes
29821 by pointer, using vector_sizes rather than a bitmask.
29822 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
29823 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
29824 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
29826 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
29827 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
29828 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
29829 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
29830 * omp-general.c (omp_max_vf): Likewise.
29831 * omp-low.c (omp_clause_aligned_alignment): Likewise.
29832 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
29833 * tree-vect-loop.c (vect_analyze_loop): Likewise.
29834 * tree-vect-slp.c (vect_slp_bb): Likewise.
29835 * doc/tm.texi: Regenerate.
29836 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
29838 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
29839 the vector size as a poly_uint64 rather than an unsigned int.
29840 (current_vector_size): Change from an unsigned int to a poly_uint64.
29841 (get_vectype_for_scalar_type): Update accordingly.
29842 * tree.h (build_truth_vector_type): Take the size and number of
29843 units as a poly_uint64 rather than an unsigned int.
29844 (build_vector_type): Add a temporary overload that takes
29845 the number of units as a poly_uint64 rather than an unsigned int.
29846 * tree.c (make_vector_type): Likewise.
29847 (build_truth_vector_type): Take the number of units as a poly_uint64
29848 rather than an unsigned int.
29850 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29851 Alan Hayward <alan.hayward@arm.com>
29852 David Sherwood <david.sherwood@arm.com>
29854 * target.def (get_mask_mode): Take the number of units and length
29855 as poly_uint64s rather than unsigned ints.
29856 * targhooks.h (default_get_mask_mode): Update accordingly.
29857 * targhooks.c (default_get_mask_mode): Likewise.
29858 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
29859 * doc/tm.texi: Regenerate.
29861 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29862 Alan Hayward <alan.hayward@arm.com>
29863 David Sherwood <david.sherwood@arm.com>
29865 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
29866 * omp-general.c (omp_max_vf): Likewise.
29867 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
29868 (expand_omp_simd): Handle polynomial safelen.
29869 * omp-low.c (omplow_simd_context): Add a default constructor.
29870 (omplow_simd_context::max_vf): Change from int to poly_uint64.
29871 (lower_rec_simd_input_clauses): Update accordingly.
29872 (lower_rec_input_clauses): Likewise.
29874 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29875 Alan Hayward <alan.hayward@arm.com>
29876 David Sherwood <david.sherwood@arm.com>
29878 * tree-vectorizer.h (vect_nunits_for_cost): New function.
29879 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
29880 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
29881 (vect_analyze_slp_cost): Likewise.
29882 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
29883 (vect_model_load_cost): Likewise.
29885 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29886 Alan Hayward <alan.hayward@arm.com>
29887 David Sherwood <david.sherwood@arm.com>
29889 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
29890 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
29891 from an unsigned int * to a poly_uint64_pod *.
29892 (calculate_unrolling_factor): New function.
29893 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
29895 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29896 Alan Hayward <alan.hayward@arm.com>
29897 David Sherwood <david.sherwood@arm.com>
29899 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
29900 from an unsigned int to a poly_uint64.
29901 (_loop_vec_info::slp_unrolling_factor): Likewise.
29902 (_loop_vec_info::vectorization_factor): Change from an int
29904 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
29905 (vect_get_num_vectors): New function.
29906 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
29907 (vect_get_num_copies): Use vect_get_num_vectors.
29908 (vect_analyze_data_ref_dependences): Change max_vf from an int *
29909 to an unsigned int *.
29910 (vect_analyze_data_refs): Change min_vf from an int * to a
29912 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
29913 than an unsigned HOST_WIDE_INT.
29914 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
29915 (vect_analyze_data_ref_dependence): Change max_vf from an int *
29916 to an unsigned int *.
29917 (vect_analyze_data_ref_dependences): Likewise.
29918 (vect_compute_data_ref_alignment): Handle polynomial vf.
29919 (vect_enhance_data_refs_alignment): Likewise.
29920 (vect_prune_runtime_alias_test_list): Likewise.
29921 (vect_shift_permute_load_chain): Likewise.
29922 (vect_supportable_dr_alignment): Likewise.
29923 (dependence_distance_ge_vf): Take the vectorization factor as a
29924 poly_uint64 rather than an unsigned HOST_WIDE_INT.
29925 (vect_analyze_data_refs): Change min_vf from an int * to a
29927 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
29928 vfm1 as a poly_uint64 rather than an int. Make the same change
29929 for the returned bound_scalar.
29930 (vect_gen_vector_loop_niters): Handle polynomial vf.
29931 (vect_do_peeling): Likewise. Update call to
29932 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
29933 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
29935 * tree-vect-loop.c (vect_determine_vectorization_factor)
29936 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
29937 (vect_get_known_peeling_cost): Likewise.
29938 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
29939 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
29940 (vect_transform_loop): Likewise. Use the lowest possible VF when
29941 updating the upper bounds of the loop.
29942 (vect_min_worthwhile_factor): Make static. Return an unsigned int
29943 rather than an int.
29944 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
29945 polynomial unroll factors.
29946 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
29947 (vect_make_slp_decision): Likewise.
29948 (vect_supported_load_permutation_p): Likewise, and polynomial
29950 (vect_analyze_slp_cost): Handle polynomial vf.
29951 (vect_slp_analyze_node_operations): Likewise.
29952 (vect_slp_analyze_bb_1): Likewise.
29953 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
29954 than an unsigned HOST_WIDE_INT.
29955 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
29956 (vectorizable_load): Handle polynomial vf.
29957 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
29959 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
29961 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29962 Alan Hayward <alan.hayward@arm.com>
29963 David Sherwood <david.sherwood@arm.com>
29965 * match.pd: Handle bit operations involving three constants
29966 and try to fold one pair.
29968 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29970 * tree-vect-loop-manip.c: Include gimple-fold.h.
29971 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
29972 niters_maybe_zero parameters. Handle other cases besides a step of 1.
29973 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
29974 Add a path that uses a step of VF instead of 1, but disable it
29976 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
29977 and niters_no_overflow parameters. Update calls to
29978 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
29979 Create a new SSA name if the latter choses to use a ste other
29980 than zero, and return it via niters_vector_mult_vf_var.
29981 * tree-vect-loop.c (vect_transform_loop): Update calls to
29982 vect_do_peeling, vect_gen_vector_loop_niters and
29983 slpeel_make_loop_iterate_ntimes.
29984 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
29985 (vect_gen_vector_loop_niters): Update declarations after above changes.
29987 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
29989 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
29990 128-bit round to integer instructions.
29991 (ceil<mode>2): Likewise.
29992 (btrunc<mode>2): Likewise.
29993 (round<mode>2): Likewise.
29995 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29997 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
29998 unaligned VSX load/store on P8/P9.
29999 (expand_block_clear): Allow the use of unaligned VSX
30000 load/store on P8/P9.
30002 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30004 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
30006 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
30007 swap associated with both a load and a store.
30009 2018-01-02 Andrew Waterman <andrew@sifive.com>
30011 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
30012 * config/riscv/riscv.md (clear_cache): Use it.
30014 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
30016 * web.c: Remove out-of-date comment.
30018 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30020 * expr.c (fixup_args_size_notes): Check that any existing
30021 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
30022 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
30023 (emit_single_push_insn): ...here.
30025 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30027 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
30028 (const_vector_encoded_nelts): New function.
30029 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
30030 (const_vector_int_elt, const_vector_elt): Declare.
30031 * emit-rtl.c (const_vector_int_elt_1): New function.
30032 (const_vector_elt): Likewise.
30033 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
30034 of CONST_VECTOR_ELT.
30036 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30038 * expr.c: Include rtx-vector-builder.h.
30039 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
30040 directly on the tree encoding.
30041 (const_vector_from_tree): Likewise.
30042 * optabs.c: Include rtx-vector-builder.h.
30043 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
30044 sequence of "u" values.
30045 * vec-perm-indices.c: Include rtx-vector-builder.h.
30046 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
30047 directly on the vec_perm_indices encoding.
30049 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30051 * doc/rtl.texi (const_vector): Describe new encoding scheme.
30052 * Makefile.in (OBJS): Add rtx-vector-builder.o.
30053 * rtx-vector-builder.h: New file.
30054 * rtx-vector-builder.c: Likewise.
30055 * rtl.h (rtx_def::u2): Add a const_vector field.
30056 (CONST_VECTOR_NPATTERNS): New macro.
30057 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
30058 (CONST_VECTOR_DUPLICATE_P): Likewise.
30059 (CONST_VECTOR_STEPPED_P): Likewise.
30060 (CONST_VECTOR_ENCODED_ELT): Likewise.
30061 (const_vec_duplicate_p): Check for a duplicated vector encoding.
30062 (unwrap_const_vec_duplicate): Likewise.
30063 (const_vec_series_p): Check for a non-duplicated vector encoding.
30064 Say that the function only returns true for integer vectors.
30065 * emit-rtl.c: Include rtx-vector-builder.h.
30066 (gen_const_vec_duplicate_1): Delete.
30067 (gen_const_vector): Call gen_const_vec_duplicate instead of
30068 gen_const_vec_duplicate_1.
30069 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
30070 (gen_const_vec_duplicate): Use rtx_vector_builder.
30071 (gen_const_vec_series): Likewise.
30072 (gen_rtx_CONST_VECTOR): Likewise.
30073 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
30074 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
30075 Build a new vector rather than modifying a CONST_VECTOR in-place.
30076 (handle_special_swappables): Update call accordingly.
30077 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
30078 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
30079 Build a new vector rather than modifying a CONST_VECTOR in-place.
30080 (handle_special_swappables): Update call accordingly.
30082 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30084 * simplify-rtx.c (simplify_const_binary_operation): Use
30085 CONST_VECTOR_ELT instead of XVECEXP.
30087 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30089 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
30090 the selector elements to be different from the data elements
30091 if the selector is a VECTOR_CST.
30092 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
30093 ssizetype for the selector.
30095 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30097 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
30098 before testing each element individually.
30099 * tree-vect-generic.c (lower_vec_perm): Likewise.
30101 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30103 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
30104 * selftest-run-tests.c (selftest::run_tests): Call it.
30105 * vector-builder.h (vector_builder::operator ==): New function.
30106 (vector_builder::operator !=): Likewise.
30107 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
30108 (vec_perm_indices::all_from_input_p): New function.
30109 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
30110 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
30111 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
30112 instead of reading the VECTOR_CST directly. Detect whether both
30113 vector inputs are the same before constructing the vec_perm_indices,
30114 and update the number of inputs argument accordingly. Use the
30115 utility functions added above. Only construct sel2 if we need to.
30117 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30119 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
30120 the broadcast of the low byte.
30121 (expand_mult_highpart): Use an explicit encoding for the permutes.
30122 * optabs-query.c (can_mult_highpart_p): Likewise.
30123 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
30124 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
30125 (vectorizable_bswap): Likewise.
30126 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
30127 explicit encoding for the power-of-2 permutes.
30128 (vect_permute_store_chain): Likewise.
30129 (vect_grouped_load_supported): Likewise.
30130 (vect_permute_load_chain): Likewise.
30132 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30134 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
30135 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
30136 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
30137 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
30138 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
30139 (vect_gen_perm_mask_any): Likewise.
30141 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30143 * int-vector-builder.h: New file.
30144 * vec-perm-indices.h: Include int-vector-builder.h.
30145 (vec_perm_indices): Redefine as an int_vector_builder.
30146 (auto_vec_perm_indices): Delete.
30147 (vec_perm_builder): Redefine as a stand-alone class.
30148 (vec_perm_indices::vec_perm_indices): New function.
30149 (vec_perm_indices::clamp): Likewise.
30150 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
30151 (vec_perm_indices::new_vector): New function.
30152 (vec_perm_indices::new_expanded_vector): Update for new
30153 vec_perm_indices class.
30154 (vec_perm_indices::rotate_inputs): New function.
30155 (vec_perm_indices::all_in_range_p): Operate directly on the
30156 encoded form, without computing elided elements.
30157 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
30158 encoding. Update for new vec_perm_indices class.
30159 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
30160 the given vec_perm_builder.
30161 (expand_vec_perm_var): Update vec_perm_builder constructor.
30162 (expand_mult_highpart): Use vec_perm_builder instead of
30163 auto_vec_perm_indices.
30164 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
30165 vec_perm_indices instead of auto_vec_perm_indices. Use a single
30166 or double series encoding as appropriate.
30167 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
30168 vec_perm_indices instead of auto_vec_perm_indices.
30169 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
30170 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
30171 (vect_permute_store_chain): Likewise.
30172 (vect_grouped_load_supported): Likewise.
30173 (vect_permute_load_chain): Likewise.
30174 (vect_shift_permute_load_chain): Likewise.
30175 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
30176 (vect_transform_slp_perm_load): Likewise.
30177 (vect_schedule_slp_instance): Likewise.
30178 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
30179 (vectorizable_mask_load_store): Likewise.
30180 (vectorizable_bswap): Likewise.
30181 (vectorizable_store): Likewise.
30182 (vectorizable_load): Likewise.
30183 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
30184 vec_perm_indices instead of auto_vec_perm_indices. Use
30185 tree_to_vec_perm_builder to read the vector from a tree.
30186 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
30187 vec_perm_builder instead of a vec_perm_indices.
30188 (have_whole_vector_shift): Use vec_perm_builder and
30189 vec_perm_indices instead of auto_vec_perm_indices. Leave the
30190 truncation to calc_vec_perm_mask_for_shift.
30191 (vect_create_epilog_for_reduction): Likewise.
30192 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
30193 from auto_vec_perm_indices to vec_perm_indices.
30194 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
30195 instead of changing individual elements.
30196 (aarch64_vectorize_vec_perm_const): Use new_vector to install
30197 the vector in d.perm.
30198 * config/arm/arm.c (expand_vec_perm_d::perm): Change
30199 from auto_vec_perm_indices to vec_perm_indices.
30200 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
30201 instead of changing individual elements.
30202 (arm_vectorize_vec_perm_const): Use new_vector to install
30203 the vector in d.perm.
30204 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
30205 Update vec_perm_builder constructor.
30206 (rs6000_expand_interleave): Likewise.
30207 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
30208 (rs6000_expand_interleave): Likewise.
30210 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30212 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
30213 to qimode could truncate the indices.
30214 * optabs.c (expand_vec_perm_var): Likewise.
30216 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30218 * Makefile.in (OBJS): Add vec-perm-indices.o.
30219 * vec-perm-indices.h: New file.
30220 * vec-perm-indices.c: Likewise.
30221 * target.h (vec_perm_indices): Replace with a forward class
30223 (auto_vec_perm_indices): Move to vec-perm-indices.h.
30224 * optabs.h: Include vec-perm-indices.h.
30225 (expand_vec_perm): Delete.
30226 (selector_fits_mode_p, expand_vec_perm_var): Declare.
30227 (expand_vec_perm_const): Declare.
30228 * target.def (vec_perm_const_ok): Replace with...
30229 (vec_perm_const): ...this new hook.
30230 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
30231 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
30232 * doc/tm.texi: Regenerate.
30233 * optabs.def (vec_perm_const): Delete.
30234 * doc/md.texi (vec_perm_const): Likewise.
30235 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
30236 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
30237 expand_vec_perm for constant permutation vectors. Assert that
30238 the mode of variable permutation vectors is the integer equivalent
30239 of the mode that is being permuted.
30240 * optabs-query.h (selector_fits_mode_p): Declare.
30241 * optabs-query.c: Include vec-perm-indices.h.
30242 (selector_fits_mode_p): New function.
30243 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
30244 is defined, instead of checking whether the vec_perm_const_optab
30245 exists. Use targetm.vectorize.vec_perm_const instead of
30246 targetm.vectorize.vec_perm_const_ok. Check whether the indices
30247 fit in the vector mode before using a variable permute.
30248 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
30249 vec_perm_indices instead of an rtx.
30250 (expand_vec_perm): Replace with...
30251 (expand_vec_perm_const): ...this new function. Take the selector
30252 as a vec_perm_indices rather than an rtx. Also take the mode of
30253 the selector. Update call to shift_amt_for_vec_perm_mask.
30254 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
30255 Use vec_perm_indices::new_expanded_vector to expand the original
30256 selector into bytes. Check whether the indices fit in the vector
30257 mode before using a variable permute.
30258 (expand_vec_perm_var): Make global.
30259 (expand_mult_highpart): Use expand_vec_perm_const.
30260 * fold-const.c: Includes vec-perm-indices.h.
30261 * tree-ssa-forwprop.c: Likewise.
30262 * tree-vect-data-refs.c: Likewise.
30263 * tree-vect-generic.c: Likewise.
30264 * tree-vect-loop.c: Likewise.
30265 * tree-vect-slp.c: Likewise.
30266 * tree-vect-stmts.c: Likewise.
30267 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
30269 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
30270 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
30271 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
30272 (aarch64_vectorize_vec_perm_const): ...this new function.
30273 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
30274 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
30275 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
30276 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
30277 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
30278 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
30279 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
30281 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
30282 check for NEON modes.
30283 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
30284 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
30285 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
30286 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
30288 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
30289 the old VEC_PERM_CONST conditions.
30290 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
30291 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
30292 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
30293 (ia64_vectorize_vec_perm_const_ok): Merge into...
30294 (ia64_vectorize_vec_perm_const): ...this new function.
30295 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
30296 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
30297 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
30298 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
30299 * config/mips/mips.c (mips_expand_vec_perm_const)
30300 (mips_vectorize_vec_perm_const_ok): Merge into...
30301 (mips_vectorize_vec_perm_const): ...this new function.
30302 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
30303 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
30304 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
30305 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
30306 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
30307 (rs6000_expand_vec_perm_const): Delete.
30308 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
30310 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
30311 (altivec_expand_vec_perm_const_le): Take each operand individually.
30312 Operate on constant selectors rather than rtxes.
30313 (altivec_expand_vec_perm_const): Likewise. Update call to
30314 altivec_expand_vec_perm_const_le.
30315 (rs6000_expand_vec_perm_const): Delete.
30316 (rs6000_vectorize_vec_perm_const_ok): Delete.
30317 (rs6000_vectorize_vec_perm_const): New function.
30318 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
30319 an element count and rtx array.
30320 (rs6000_expand_extract_even): Update call accordingly.
30321 (rs6000_expand_interleave): Likewise.
30322 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
30323 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
30324 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
30325 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
30326 (rs6000_expand_vec_perm_const): Delete.
30327 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
30328 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
30329 (altivec_expand_vec_perm_const_le): Take each operand individually.
30330 Operate on constant selectors rather than rtxes.
30331 (altivec_expand_vec_perm_const): Likewise. Update call to
30332 altivec_expand_vec_perm_const_le.
30333 (rs6000_expand_vec_perm_const): Delete.
30334 (rs6000_vectorize_vec_perm_const_ok): Delete.
30335 (rs6000_vectorize_vec_perm_const): New function. Remove stray
30336 reference to the SPE evmerge intructions.
30337 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
30338 an element count and rtx array.
30339 (rs6000_expand_extract_even): Update call accordingly.
30340 (rs6000_expand_interleave): Likewise.
30341 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
30342 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
30344 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
30346 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30348 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
30349 vector mode and that that mode matches the mode of the data
30351 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
30352 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
30353 directly using expand_vec_perm_1 when forcing selectors into
30355 (expand_vec_perm_var): New function, split out from expand_vec_perm.
30357 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30359 * optabs-query.h (can_vec_perm_p): Delete.
30360 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
30361 * optabs-query.c (can_vec_perm_p): Split into...
30362 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
30363 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
30364 particular selector is valid.
30365 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
30366 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
30367 (vect_grouped_load_supported): Likewise.
30368 (vect_shift_permute_load_chain): Likewise.
30369 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
30370 (vect_transform_slp_perm_load): Likewise.
30371 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
30372 (vectorizable_bswap): Likewise.
30373 (vect_gen_perm_mask_checked): Likewise.
30374 * fold-const.c (fold_ternary_loc): Likewise. Don't take
30375 implementations of variable permutation vectors into account
30376 when deciding which selector to use.
30377 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
30378 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
30379 with a false third argument.
30380 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
30381 to test whether the constant selector is valid and can_vec_perm_var_p
30382 to test whether a variable selector is valid.
30384 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30386 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
30387 * optabs-query.c (can_vec_perm_p): Likewise.
30388 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
30389 instead of vec_perm_indices.
30390 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
30391 (vect_gen_perm_mask_checked): Likewise,
30392 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
30393 (vect_gen_perm_mask_checked): Likewise,
30395 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30397 * optabs-query.h (qimode_for_vec_perm): Declare.
30398 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
30399 (qimode_for_vec_perm): ...this new function.
30400 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
30402 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30404 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
30405 does not have a conditional at the top.
30407 2018-01-02 Richard Biener <rguenther@suse.de>
30409 * ipa-inline.c (big_speedup_p): Fix expression.
30411 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
30414 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
30417 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
30421 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
30422 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
30423 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
30424 cond_taken_branch_cost 3->4.
30426 2018-01-01 Jakub Jelinek <jakub@redhat.com>
30428 PR tree-optimization/83581
30429 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
30430 TODO_cleanup_cfg if any changes have been made.
30432 PR middle-end/83608
30433 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
30434 convert_modes if target mode has the right side, but different mode
30437 PR middle-end/83609
30438 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
30439 last argument when extracting from CONCAT. If either from_real or
30440 from_imag is NULL, use expansion through memory. If result is not
30441 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
30442 the parts directly to inner mode, if even that fails, use expansion
30445 PR middle-end/83623
30446 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
30447 check for bswap in mode rather than HImode and use that in expand_unop
30450 Copyright (C) 2018 Free Software Foundation, Inc.
30452 Copying and distribution of this file, with or without modification,
30453 are permitted in any medium without royalty provided the copyright
30454 notice and this notice are preserved.